@@coso2 right, 2.5/5gb isn’t going to get me through 10 years (how I see the life of a switch). At this price point, if I need to buy something mid-way I might as well just go to 10Gb now.
@@coso2 99% of consumers barely use 1GB from there devices so internal networks for most home users or SOHO still don't need 2.5/5Gbps yet, but, companies need to make money so they will sell this as a stop gap and rake in the profits.
There are cheaper options. QNAP makes a 5-port 2.5Gbe switch (the QSW-1105-5T) that has an MSRP of $100 USD (though real-world prices are more like $110).
@@guspaz It's cheaper, but still pricey. Street price for that in the UK is £110 (£22/port, about right for USD -> GBP + VAT); meanwhile I can get a TP-LINK LS108G for £16.
@@melgross I jumped into the GbE @ home world early on when 8 port switches needed a full U with active cooling and finally snuck under $1000. Just a few years later, they were half the size, 1/4 the power consumption, passive and $100. Now they're plastic and $25. 10GbT has stubbornly held its per-port price for damn near the last decade that I've been watching - over that same period 1g dropped 95%. Introducing 5g and 2.5g at "most-of" and "a-big-slab-of" the 10g port pricing doesn't make an especially convincing argument for.
@@mattd5136 while it may be hard to believe, Apple was credited for bringing the cost of both 100Mb and 1Gb down when they put ports into all of their machines. A year after they did that, cards dropped from about $750 to $100, and a year later, to below $50. But Apple didn’t do that with either 2.5 or 10. They put 10 on their highest end, but lowest selling machines. What’s needed is for a major computer manufacturer to make this standard. But no one has been interested. Unless this reaches major sales. Prices will just come down very slowly.
I'd love if motherboards, desktops, and other devices would start standardizing on 2.5 GbE as the baseline (whereas 1 GbE is now), but still, these switches are 4-5x more expensive than 1G units. Until that changes I'd still rather stick with hybrid gear like QNAP's 10/2.5 Gbps switch, so it's more future-proof for when I have more 10G-capable devices. 2.5 Gbps is very appealing for me since I already have 8 rooms in my house wired with Cat5e, and pulling/rewiring to Cat6a or Cat7 is going to be annoying.
Have you tested the existing cat5e runs with 10 GbE? If they aren't tens of meters long, it would likely work well. But of course it might be unreliable if the signal is stretched to its limits. You will definitely need cat6 to run the full 100 meters of 10 GbE. But at short distances anything that resembles an ethernet cable likely works just fine.
I agree with Jeff Geerling that the QNAP QSW-M2108-2C is the more attractive solution to me -- which is odd because just yesterday I was shopping for one on Amazon and reading a review about it there from a Jeff Geerling (and a week before that I was on Leanpub downloading Ansible books from a Jeff Geerling). Anyway, I'd say 10 GbE uplinks on a 2.5 GbE switch are pretty critical here, as switches like the Mikrotik CRS309-1G-8S+IN make it pretty easy to have a (silent) home 10 GbE core nowadays, and a 10/2.5 GbE switch is a nice extension to that for the growing number of 2.5 GbE equipped gear. I'm not sure if a standalone 2.5 GbE network has much of a value proposition to me, especially not at the price discrepancy you note. The question is: what are you connecting to at that speed? A 2.5 Gbps Internet uplink (I wish!), a local server file share (more probably), etc.? I'd say that these days, working off a network file share at 1 GbE is just unworkable, and probably 2.5 GbE isn't enough of a step up. Local direct attached storage nowadays is commonly PCIe 3.0 NVMe, and increasingly PCIe 4.0 NVMe, so working over 1 GbE/2.5 GbE (essentially mechanical SATA drive speeds) just feels like your computer is freezing/broken. Compared to NVMe, even 10 GbE seem insufficient, clocking in at RAID 0 SATA 3 SSD speeds. Really, we're approaching the point where 25 GbE is the lowest bar to keep the network competitive with local storage. What's more, 30m runs of OM4 fiber optics can be had for like $50 bucks. The justification for settling for slow copper network speed is getting slim. I'd sooner pay more for more PCIe lanes on a workstation (to have space to add 25 GbE+, RDMA capable NICs) and a silent home 25 GbE (even just 8 port) switch.
@@ServeTheHomeVideo didn´t you see the 4x Intel lan cards from QNap that cost "near to nothing" .. make your own switch : QXG-2G4T-I225 (80€ here in germany)
Honestly, as a home user, I think 10USD per port will be my switching point. Currently, the only situation that saturates the current 1Gbps connection is bulk file transfers, which does not happen very often. I think what I care more about is consistent processing latency and stability.
@@tommihommi1 A competitively priced Mikrotik switch with 8-24 copper 10GB ports (with 1/2.5/5 fallback), and two (Q)SPF+/28 25GB, 40GB or 100GB uplinks would be an instant buy from me. I don't even need POE.
$22 per port and they can't put the leds in the RJ-45 ports. The LED placement is an important issue. I totally agree with you on pricing. My guess is that we are just a year or two early for low prices. Thanks for the great videos.
I thought the LED placement was nit-picking until I saw the price. These switches aren't going to age well as far as appearance. In one or two years they're going to look like the $25 bottom-of-the barrel switches we see today.
Agree that the per-port cost for 2.5g is quite a jump from 1g. It's high enough that I just went with 1g or forked out for 10g for the stuff that benefited from it for personal use. This would be perfect with an SPF+ port or 10g for uplink but without the £400 tag of the TEG-30102WS. Prices and product choice is slowly getting better but there's still some way to go yet.
$100 for 8 ports would be ideal (~$13/port), but $120 is probably the most I'd go. Let's be real, these will mostly be used as edge switches, not backbone switches. Since most residential internet connections in the US won't see even gigabit speeds for a significant period of time, and a gigabit switch with 10Gig uplink can be had for the current asking price of the 8 port unit ($189), I think they make very little sense at the current price point for home users like myself. Unrelated; as much as I like to geek out about crazy high end equipment I can never own/have no use for, this kind of content (consumer/prosumer networking gear like unmanaged switches) is what I really need. Keep up the great work!
Glad they are finally here, however I feel like the benefit of going from 1gbe to 2.5g is too costly for enthusiasts in most cases. Meaning it makes more sense to go 10gig if going out and buying bespoke hardware. 2.5g needs to replace 1gbe as the default, with 10gig still the enthusiasts option I feel
I questioned that myself. I mean, we went from 10Mb Ethernet to 100Mb, and then from 100Mb to 1Gb, and then from 1Gb to 10Gb. Every jump was a 10x increase. But going from 1Gb to 2.5Gb... I doubt in most cases you could even tell the different. Like going from 100Mb to 200Mb. Not worth buying new equipment.
@@eosjoe565 What I don't understand is why 2.5GB pops up on all the new Intel motherboards? I mean you can buy USB 5GB NICs that are faster! Feels like penny pinching to me.
@@kwinzman You need to be careful with these 5GB USB NICs. A lot of them actually are not faster than 2.5GB. Seems to be a weird rip-off, but you should do your research before purchasing.
@@fbnx4219 I know that the qna-uc5g1t for example which links at 5GBE can only do 3.5Gbit/s because it's limited by the 5Gbit/s USB3.0. But it acts as as a dirt cheap upgrade at 80 bucks, usable in pretty much any machine manufactured in the last few years, without even opening it. There is no risk of blue screening the computer when suddenly unplugging like with some Thunderbolt 3 NICs. I suspect this problem will go away entirely with USB4. And then 2.5GBE on the motherboard will look even more outdated and peny-pinching cheap than it does already.
Enthusiasts got second hand fiber optical 10GbE gear years ago and don't have to care about this. I know I did. Let these switch manufacturers milk the early adopters of this new market for all I care.
I mean, if you absolutely need the cable lengh of RJ45, these switches maybe make sense, but the 4x speed increase of the MikroTik 10Gbit switch with SFP+ ports(at the same price but with only four 10Gbit and one 1Gbit port) seem to make much more sense, if you can satisfy the bandwidth of course
At it's current pricing, per port cost is way too expensive for regular consumers who just want to plug and play. I agree with you on the recommended pricing.
I think I would be buying some of these if they were a third of their actual price. Probably it will be 1-2 years from now, but I am quite confident that we will get there.
Mikrotik can make a 4x10GbE switch for U$S149, an 8x10GbE for U$S269 and the one that I bought the 16 port 10GbE with redundant power supplies for U$S399. And TRENDnet expects U$S180 for an unmanaged 8x2.5Gbe switch? I know, different customer base, but my point is... how big is the margin TRENDnet gets on these things? IMHO motherboard vendors were never interested in expending a single cent more than necessary because of the low margins they manage (not long ago having an Intel 1GbE instead of a Realtek based solution was the "high end" solution on expensive motherboards), and the lack on interest on the average customer never pushed them to do anything else. There are LOTS of marketing and money invested on WiFi technology, which before WiFi6 it was really "meh" technical wise, but the average customer wanted that... no cables, big numbers. It's a pity, really, the enthusiast has no choice but used enterprise NICs and this switch will be a market flop. It targets a market that... doesn't exists. An unmanaged 4/5 2.5GbE switch should be U$S50, and should be the norm. I updated my home network to 1Gbps... maybe 15 years ago. Sorry for the rant. It's a shame we are not using 10GbE as a cheap standard and replacement for 1GbE like 1GbE was to Fast Ethernet.
Still blows my mind that 15yrs ago I bought a netgear 8 port 1Gbps unmanaged switch for $75. Really wish I knew what the hold up is getting faster networking down to affordable prices. 2006 me was convinced by at least 2015 10GbE would be in the home. I was wrong.
Excellent points, Patrick! I just spent the last few days doing a "drop-in" replacement of a Linksys 8-port Gigabit switch with an 8-port TP-LINK 2.5GbE switch. That new 2.5G switch caused zero problems. All the problems we encountered happened with 4 x USB 3.0 to 2.5GbE ethernet adapters, and 1 x 2.5GbE PCIe x1 add-in card -- trying 5 different vendors. Because we still use Windows XP on our backup storage servers, we naturally had hoped the USB 3.0 dongles would work on those PCs. NOT!! Yes, I already hear you saying that XP is obsolete. HOWEVER, the Realtek website does list a download option for "WinXP Auto Installation Program" under the RTL8156 and RTL8156B controllers. And, despite my good faith efforts to request Tech Support (now that I am a paying customer), what few responses I did receive from the sellers were terribly inadequate. The email address for Tech Support at asustor is "over quota" and my question to them is still waiting for answers. I haven't told StarTech yet, but their dongle starts up at 100Mbps, and must be DISABLED and ENABLED in Network Connections to change the speed to 2.5Gbps. Also, that dongle required a hybrid USB cable that supplies extra power from a USB 2.0 port: the latter problem may be due to limited power available from the USB 3.0 ports at the real I/O panel of our HP Z220 tower workstation. So, a hybrid USB cable fixed that one fault. The good news is that all 5 of those adapters installed mostly OK on Windows 7 and Windows 10 PCs. And, the excellent software at drivereasy.com made it relatively simple to upgrade device drivers to the latest versions. If you're not already aware of the latter website, I have been updating drivers with their Windows program for several months now, and I have not had one single problem with any driver updates. I should add that our new TP-LINK switch is on the floor, below one workstation monitor, and we point a DC fan directly at that switch because it does get warm to the touch. Hope this helps. p.s. Yes, Patrick, your astute analysis of prices is RIGHT ON THE MONEY: prices should naturally go downwards as awareness of 2.5GbE expands worldwide.
I should add that performance is what we expected, when other overhead sources are taken into account. For example, by watching the Ethernet port activity with Windows 10 Task Manager, the real-time graphs do approach a 2.5-to-1.0 ratio with averages of 10-to-4, more or less. This ratio is most obvious when we run XCOPY twice in a row, and file properties are mostly memory-resident during the second run of that Windows utility. Of course, if a Windows 10 PC uses a 2.5G controller to talk with a Windows 7 PC with a 1.0G controller, the overall speed is reduced by the latter "weak link" in that chain. What impressed me the most was the obvious speed jumps we saw when copying large driver software folders between 2.5G ports, particularly when both PCs were cabled directly to the TP-LINK switch, and did not need to traverse any switch "cascades" or pass traffic thru our router. With that experience under our belts, we're now planning to upgrade to TP-LINK model TL-NG421 PCIe x1 add-in cards, and keep the dongles in spare parts inventory, for now. We were delighted to discover that the latter add-in card also works in a PCIe Gen1 x1 expansion slot i.e. same raw bandwidth of 2.5G per x1 lane. Last thing: I'll be contacting our router vendor to ask if they have any plans to offer a drop-in 2.5G replacement for our 4-port Gigabit router.
This is great, although I wish it came with a one or two 10gbe ports.. It is still an "affordable" switch for what it offers. Its nice to see someone taking the time to review network gear -- This is exactly why I come to this channel -- Thank you, Patrick! :)
For me personally, I'd love to upgrade my small homelab to something faster than 1Gb, but for it to be really feasible for me, I think things would have to hit the $10-12 per port range or
For a home what I am looking for is 18 ports. 4 x 2.5gbe poe for wifi, 4 x 1gbe poe for cameras, 4 x 2.5gbe for desktops and nuc, 4x 1gbe for tvs and other legacy devices, and 2 x 10gbe for servers/ Nas. Maybe a 26 port that you double the nonPOE ports.
I think $10/port gives them enough of a profit over the $3/port for current unmanaged switches and still feels like a fair price as a user. Especially since OEM's are now including 2.5Gb Ethernet as standard equipment. It's no longer a special order item like a 10/40GB device.
@@thecockflock At the time of writing this, All Qnap has some, but for $124 instead of $110. My point is that it's still cheaper than the $140 for the TEG-S350.
After looking for years for faster than 1GbE solutions, I finally made a jump to 10GbE over fibre. Cards are super cheap on eBay and 2 cheap Chinese SFP+ modules + fibre patch cable is cheaper than a DAC cable or a pair of -T modules.
You see it the same way that I do ... 11:38 ... "That gap is too much". I agree. That's why I don't think that 2.5Gb will take off. Most people rather spend a little more and get 10Gb. At the price point of $24/port, that is just stupid and ridiculous. Way too much to make it universally acceptable in its pricing right now. Basically, 10G is gonna be accepted faster due to these high prices.
If you look at the history of consumer and some business hardware, (here in the UK), you will see that even in the first decade of this century motherboards, until later in the decade were still at 100 mbps, though Laptops/notebooks had gone up to a Gig. That meant a lot of places still had 100 MB infrastructure with 1 Gig uplinks. It was only as more items got 1 gig that the market moved. So before the end of that decades where I worked we had gone to 1 Gig throughout with a couple of 10 Gig links, but hey were expensive. You found alternative way to increase uplinks, etc bandwidth if required. These days I only run a small home lab and that is still on 1 Gig and is OK for my current requirements though will look at going higher when the prices significantly reduce, or my needs change. The benefit of 2.5 gbps is that I can continue most likely to use my CAT5E cabling, and I have boxes of that in the shed, even maybe be able to handle 5 gig at a pinch. However what would be interesting is due to reduced costs of production as sales increase and prices come down will there then be that much difference in costs between a 2.5 and 5 gig switch on a per port basis. Incidentally all my switches are metal cased. The only item that isn't is my router that connects via copper to FTTC. Regarding the port indicating LEDs. I do agree that above the port is best. Where I worked we had racks of switches and most indicators were above the port and they were so easy to detect that they were fully operational, and at what bandwidth , than the few that were to the side of the ports.
For my home network I would struggle to spend much more on a 2.5gb switch than 1gb. The way I'm thinking about it is the use case for every device needing >1gb would be if I have an internet connection that can saturate it, but I don't so realistically there are only a few devices passing around large amounts of local traffic. If the amount of traffic is noticeably slow on a 1gb connection then at that point I'm just going to move those devices to 10gb.
For an 2.5GbE 8 port unmanaged switch around 100USD is what I would be willing to spend. Around the 200USD mark I would rather just hop on ebay or similar sites and pick up a used managed switch. But to be completely honest I wish companies would just skip 2.5GbE and go right to 5 or even 10GbE.
@@kwinzman i don't mind intel doing this as much, as we can add network cards. Would be better if they went to 5GBE, but at least they are leaving the GB phase. However, we hardly have any option for switches at 5GBE (or more) that does not cost an arm and a leg... I would have an easier time paying more overall for a cheaper 5GBE switch + adding cards on each machine (but not all at the same time, perhaps 1 or 2 each month) than having 5GBe on all machines from the start, but paying a high price upfront for the switch... Obviously would be better if 5Gbe or more were the intel move AND switches were cheap :)
@@AudreyRobinel And even rarer is to see an actual 25GBE SFP28 uplink or ideally 2 SFP28 uplinks on your 5/10GBE access switch. For something that would actually make sense on the switch side you suddenly pay 100 times more than for similar 1GBE switches with SFP+ uplinks.
Heck. Mikrotik's 10G offerings (4+1, and 8+1) at $140 and $256 respectively... managed, fanless... very competitive as well, as long as you're buying DAC cables and not having to invest into SFP modules. It's good to see new 2.5G offerings, but I agree the price needs to come down further.
Your pricing (16:50) is somewhat right. They should charge double say what 1Gb switches are, say $55 for FIVE 2.5Gb ports and then (maybe) $85 / $90 for EIGHT 2.5Gb ports. If not, I'd buy 10G switch instead. It has to be HALF what they are charging now.
I recently purchased a Ubiquiti Enterprise 24 PoE for my home with 12x1Gb, 12x2.5Gb, 2x10Gb SFP+ ports, at $800. I love it because it supports Comcast's 1.2Gb internet at full speed, and has PoE for security cameras and APs. Not all of the ports are 2.5Gb, but for the security cameras and non-core equipment that's not a concern. It seemed like a good compromise, especially for the PoE.
I think if you go over $100 it has to have compelling features such as a 10Gb uplink or be managed. An unmanaged 8 port 2.5 with a single 10 gig (sfp+) at around 150 would be right around where I would be ok with it. I would prefer it nicely in the 120-140 range. A 5 port unmanaged 2.5gb switch needs to be under $75 for general adoption or it needs to have a higher uplink to be any more expensive.
Honestly, at this price it’s still too much. For $140 I’d rather do the tiny mikrotik with SFP+. $23/port is about 3 times as much as what it’s worth. You mentioned that a gig port is $3 so makes sense that a 2.5G port should go around 2.5x$3=$7.5/port. Anything above $10 a port makes no sense.
Im pretty much at the same page as you are talking about, i want a 2.5gb switch, but they are far to expensive atm to make it an option, versus just sticking to a 1gbit switch, it is faster sure, but for my use, it wouldnt make a hell of difference, at least to a point where a 10x price premium is worth it. To be relevant pricewise it would have to come down to something like $60-$70 for a 5 port switch.
I don’t want to spend over $100 for a 4/5 port unmanaged switch. At that size its just going to be home use plugged in and forgotten about. That comes out to 1 uplink and $25 per device added to the wired network.
With 1G switches at $2 - $3 per port, going to 10x the price for 2.5G is totally unreasonable. Yes, they’ll definitely sell some at those prices, but sales volumes won’t go up until they get it down to around $10 per port. I’m pretty sure for most consumers, 2.5 times the performance is just not worth much more than 2 or 3 times the price.
I agree to 100% , and this as someone who do e the change to 10gbit 2 years ago. That might be different for 5gbit, as even with 10gbit you have to decide if you rather do the step to SFP+ anyway. The switch manufacturer just missed the opportunity to really make a step to 5gbit and offer a real compromise for consumer. With the same cabling able to handle the 5x throughput.
Edit: Also, I absolutely won't be buying any of these until I can get a 5-port switch for $60 max. I think your $12/port assessment makes perfect sense. I completely agree with your thoughts on the price jump from 1Gbe to 2.5Gbe. I don't really see the justification...it doesn't seem like the parts or engineering work are dramatically different from 1Gbe. So ... it seems like the prices make no sense. It should be *way* cheaper. Like less than half lol
Unmanaged means one does not have a management interface on the switch to set VLANs there. One can still set VLANs on other switches/ NICs in the infrastructure.
If by "supports vlan" you mean passes the vlan tag through unmodified then all dumb layer 2 switches "support vlan". The switches that don't are using a switch chip that can handle tagging and untagging vlans but it's just configured improperly and being unmanaged you can change it from how the manufacturer configured it. Likewise some switch chips out there support STP but can be configured improperly to effectively just eat BPDU messages and forward all traffic thus breaking STP for those ports when hooked up to a switch that supports STP. As far as a dumb layer 2 switch is concerned, 802.1q frames are valid regular ethernet frames. The source and destination mac addresses are in the right spot, the first 16 bits of the 802.1q tag are 0x8100 so as far as the switch is concerned, this is the ethertype field to pass along unmodified. It extends the max length of a frame, but modern switch chips generally all support larger frames anyways.
@@ServeTheHomeVideo Just a few weeks ago I read about a switch that was unmanaged but had a dip switch that when enabled put each port in a separate VLAN. A switch like that with a 10G uplink (and preferably 2.5G ports) would be awesome.
@@ServeTheHomeVideo sure, it's that some tp-link switches has a "management" ip that you can access and do Vlan tagging, but they are still considered unmanaged. Tp-link used to call them Smart Switch... Idk if I'm wrong but for SOHO just being able to do Vlan tagging is great!
@@AndrewMerts I expressed it poorly. What I meant was that I've seen some SMB switches that are not managed but can still do Vlan tagging. I used to have a SG108e from tp-link that could do Vlan tagging, was not a managed switch, and it couldn't do LACP for example.
My current problem with 2.5GBE is that while it is faster than GBE, it is not super fast. It is aproximately good hard drive fast, or a bit faster. But it is way slower than even sata ssd speeds, and let's not even talk about nvme. So going 2.5GBE for me is already a compromise, i settle for lower speed that what i would want. I could be ok with that if it were cheap, but it is still quite expensive... So i feel like i am accepting "MEH" performances and still paying premium... I would feel more satisfied by paying twice as much, but for 10GB lan.
The QSW-308S is an 8-port gigabit switch with three 10 gigabit SFP+ ports for $139 USD. They have an excessive number of variants of it based on if it's managed or not, or how many SFP+/10 gig copper ports it has. Add management and a fourth SFP+ port for $40, or add a secondary 2.5/5/10 gig RJ45 port to an SFP+ for $30 a pop (it's cheaper than a copper 10 gig SFP+ module). They range from $139 for that cheapest unmanaged one (QSW-308S), to $299 for the managed one with four SFP+ and four 10 gig copper ports (QSW-M408-4C) or the one with eight 2.5 gig ports and two SFP+/copper 10 gig ports (QSW-M2108-2C). Microtik also has a $99 switch that's an 8-port GigE with two SFP+ ports.
@@guspaz Hello! this is a really interesting reference! I currently own the mikrotik 99$ switch (and in retrospect, i should have bought the larger version, with 24 gb lan and still 2 SFP+ for 140$). The QNAP offer is really interesting, for now 3SFP+ would totally do it for me... and with four i'd be set for my next phase (a TrueNAS box, while keeping the previous OMV nas). 300$ is a bit more that what i'd like to pay, however, for 8 10Gb ports, copper and SFP+, this is really nice. If i'm in at 200$ perhaps adding 50% more and getting full 10Gb is the way to go. Plus copper 10Gb means that i don't have to buy cages. However 10GbE cards seems more expensive than 10Gb SFP+ cards.
@@AudreyRobinel If you already have the $99 microtik one, you may want to simply buy their 4xSFP+ switch ($150, CRS305-1G-4S+IN) and connect it to your existing switch, either with a 1 gig uplink via the 305's gigabit copper port, or via one of the SFP+ ports with a direct attach cable. However, that would only get you two additional SFP+ ports over what you have now, due to the two ports consumed by the uplink connection. The $270 CRS309 solves that problem by having 8xSFP+ ports, but now the cost is starting to get out of control.
@@guspaz that was the plan! with both, i'd have a total of 4 available ports (2+4 -2 for uplink). But i have been looking at the 8 ports variant too, since if i'm in for 150$, i may well add 120 more and be set for a long time (i can see how i fill 4 SFP+ ports, but i don't have plans that would fill 8 for quite some time. So nice headroom for my use case here!) I have also taken note of their 24 SFP+ switch, with 2 40Gb links :) Way too expensive for what i am doing, but at 500$, this seems like an amazing per port value, and 40Gb links is something i didn't even imagine that would ever be in my price range. the 40Gb link would be nice for a SSD cache on the NAS :D (but then i need to add a pool of fast NVME drives, a 40Gb lan card to the NAS, probably a beefier CPU, more ram, etc... gets out of hand quickly ^^)
Same thing happened when the transition on the desktop from 10/100 to 1000Base-T occurred. Will probably be another year or two and we'll see 2.5G at less than $100; nice that they're fanless. So many of the first cheap gig-E switches that would fail from overheating due to crappy or clogged fans. Back then it was a little different in that a home NAS was not anywhere near as common. The upshot of this is that these switches with no higher-speed uplink ports are really only for the smallest and most basic networks while the one's like you mention with dual sfp+ ports now have a much bigger market. A managed switch with fiber capable 10G ports for a ~$250 is a steal. Make the jump people!
20 years ago I wired a house for 10/100. That infrastructure still works for 1 GbE. I might be able to get 1 Gbit Internet here, but I don't feel all that constrained by the ~100 Mbit I usually get. My currently router is limited to 1 GbE. I can bond ports on the back side to get a bit more bandwidth between the house network and my home lab, but for that use 1 GbE is good enough. I will eventually need to find a way to get 2.5 GbE to the router from the WLAN, but not until I upgrade the APs. The key problem for 2.5 GbE for me is that it really only makes sense if you are regularly saturating 1 GbE, and if that case, you are probably going to want 10 GbE (or better) anyway.
I agree that the $100-120 range is the point at which these go from being "early-adopter" to "worth considering." It'd still be great to see them get below 3 figures, though, then things like that QNAP with the 10GbE uplink ports won't seem so attractively priced. Motherboards that include 2.5GbE still tend to be on the higher end of the scale, roughly $150 and up, so there's still some of that early adopter premium in play. It's still good to see the new NUCs adopting it. My B550 board has 1x1GbE & 1x2.5GbE, so I've been tempted to try it out. Plan A was to get an 8-port 2.5GbE switch, for future expension, and add a PCIe card to my Home Lab rig (currently just one system with a far-too-loud PSU). After spending ages looking for an attractive 2.5GbE switch, I gave up and ended up buying a Netgear GS108Ev3 for about $30 to play with the switch management features (I've only ever had unmanaged switches). Plan B is to stick with 1GbE for the rest of the network and have a direct 2.5GbE link between the home lab server and main PC. Still, finding a 2.5GbE card that's reasonably priced (and doesn't look like it'll break if you look at it funny) is proving difficult.
Most home users won't need 2.5gbps and the power users have already gone to 10gbps. 2.5/5gbps equipment is DOA. Most systems are coming with dual 1gbps ports and you can bond ports and be way more cost effective. You can get dual port USB3 adapters for cheap.
I agree. I wouldn't consider buying into 2.5G until I can find a 8 port below 100$. It's not like 2.5G is for enthusiast when we even have ports on NUC systems, and for mainstream adoption the 100$ barrier is kind of a big deal.
I remember when 2.88 MB floppy disks came out but then disappeared without widespread adoption because folks were not willing to pay double. I'd pay $50 for a 5-port switch, but I'd have little interest in upgrading whilst my internet connection speed is capped at 200 Mbps.
I stumbled over (didn't try yet but think about getting) the Zyxel XGS1010-12 which I think is another interesting unmanaged switch - same price range but it enables interconnecting 1gbit, 2,5gbit and 10gbit sfp+
Not sure if I could go fully unmanaged so bought the zyxel xgs1210 so I could get 10 g uplinks with couple of 2.5 gig ports, with vlan support for about the same price as these.
This was a fantastic review however it should be around 10 per port this is overpriced. So for me to upgrade I really want a 16 port managed version of this switch for around 160 to 180 max. So in short I see zero value at the current pricing IMO you should just purchase a 10 gig switch instead of wasting money on an overpriced 2.5 switch.
I'm doing switch upgrades for our whole small office network this year. The killer product for me would be a 48-port layer-2 1u multigig switch with SFP+ (at least) uplinks and no PoE for about $1000. It somehow seems ridiculous to me that I'm pretty much stuck with 1g access ports in 2021, without spending at least $4k/switch, where I can get the equivalent 1g switch for about $500. And the office isn't quite small enough to run off just one 48-port model...
@@bjornSE That is a good deal, but doesn't work as an access switch - at least not without $40/port worth of SFP+ modules. Seen some good deals out of Mikrotik, but I've also heard of a decent rate of software problems on their switches, so not something I'd bet my reputation on quite yet.
The market for 2.5 Gb is very limited in the consumer market. Wifi has improved significantly and for the wire folks, the logical upgrade path from 1 Gb is 10 Gb.
I think I'd not buy a 2.5G switch. Either I can use the speed, then I'd get a 10G / 25G / 100G switch or speed isn't critical and there I'll just stick to 1G which is still plenty for Internet connectivity or accessing local services on the network from a desktop. But maybe other people actually do have use-cases for it.
Sound reasoning, for those with a need for quick access to large amounts of data 10Gb should have been a no-brainer years ago. Or we're in the Facebook/UA-cam/Netflix crowd and for those 1Gb is more than enough with even a very high-bitrate stream from one of the mentioned services less than 100 Mb/s.
Personally, I think 2.5Gbps is a waste of everyone's time and money. Just go straight to 10Gbps. The fact that even SATA SSDs are bottlenecked by it makes it feel old on arrival.
No-one has created at 10GbE copper device that works on Cat6 @ 100m, I don't think? People don't want to change all the cables in walls, that's a big driver for 2.5GbE. Along the same lines: WAPs that push > 1Gbps and use PoE.
@@SuperSpecies Cat 6A supports 100m 10GBE and it's not that expensive. If you have legacy cables that are really pushing the limit on the length you can always auto-negotiate NBASE-T on the 10GBE NIC. 2.5GB is penny pinching. Give my clients NICs that support up to 10GBE by default.
@@kwinzman many installations have cat 5e and cat 6 installed, and replacing the wiring in the wall with cat 6a is a massive capital expenditure and logistical nightmare at some places. Sure a new site go with cat 6a or better, no doubt.
I went straight to 10G on used hardware since it was cheaper than buying new 2.5G. If it's gonna take off I feel like 2.5G needs to beet the used market
It will never do . Its there for peopl that have no clue at all. You also find 10G overpriced consumer stuff. 2.5G existed for companys that didnt want to replace there old cabels but still go a bit faster . 10G is a sweat spot in my opinion
Maybe for 802.3bz 5 gig ports, but really the standard for Home /SMB should be ten gig to really justify the expense of upgrading the capacity of an entire network. Right now I use 1 Gig due to price and the fact that my home network uses L3 switches as I run IS-IS routing protocol and dual stack IPV4/IPV6 for all my network devices. I seriously doubt I am a typical use case though.
We should be pushing to 10G and not wasting time on a stop-gap 2.5g. Protocol that won't really live long enough to make it into mainstream devices. It's like a 250mbps port instead of a1G
I'm currently looking at an upgrade project were I would like to add a 2x port 2.5Gbps Ethernet NIC to a small group of workstations and bond the channels for a maximum transfer rate of 5Gbps. If the price per port was in the 12 - 13 dollar range then the cost of two switches and the NICs would be acceptable.
The price point of 2.3Gb Switches is part of why I didn't worry about getting 2.5Gb Ethernet on my motherboard when I built this machine a year ago. Switch prices are just too high to justify using it yet for me.
Sure, in a fantasy world they wouldn't cost anymore than 2.5x the 1Gbps, but I also realize from the manufacturers perspective there are economic and efficiency reasons that may not be entirely feasible. I also tend to go out of my way and spend a little more on 1Gbps un-managed switches to get things like a metal frame, maybe some additional mounting options, and possibly a little more reliability in the hardware given the price. Before I heard you throw out your number, I was thinking around 4x what the cheap 1Gbps switches normally cost (with a little less margin for the type of switch I usually go for), which ended up being in the same ballpark as yours.
Waiting for a 20 port with 4xSFP+10 for ~$300 to replace my trendnet 20+4SFP+ that i bought several years ago for ~$250 I too dont like the seperate pannel for LED activity/link state
To start with paying double the 1GbE equivalent is the pass fail line for most of my casual use acquaintances, but over time that has to come down. For the more advanced 2.5GbE isn't worth the hassle and they're looking to 10GbE, even if they have to go used.
I would think of that high cost as a so-called "early buyers tax", since I still find 2.5GbE a novelty. But they will definitely come down in price, just when it will would be a different and relatively unanswerable question at this point. Where it makes sense now is if you have a home network storage, where the higher network connectivity bandwidth means faster data transfers. But once the Internet side of things transition to 2.5GbE like modems, optical network terminals and routers, the prices might go down further as manufacturers push out more products and more consumers gain access to higher speeds.
Yes, exactly. As I have commented before this unmanged switch without faster uplinks worth maximum $20. Maybe 10-20% more than an equivalent 1G switch. Until price becomes resonable it is just milking uninformed consumers that don't know about 10GBE.
@@kwinzman yep, many would be better off buying 10GbE, would probably cost the same and yet future-proofing. My current ISP even offers 10Gbps plans over fibre for a cool $170 SGD. But only a few people would be able to saturate the connectivity, since even I could run on about 300Mbps just fine as I am stuck on using powerline.
@@kancheongspidergaming Interresting perspective! Personally my use case is NAS. Even a cheap SATA SSD is already too much for 2.5G Ethernet, 10G is the logical upgrade from 1GBE. But 2.5G is getting pushed by big Intel now so it might survive if it's the same price as old 1G.
On unmanaged switches we can basically do throughput, we could do pps. On higher-end switches, one tends to have a CPU, memory, and storage running Linux so perhaps one could run a database benchmark on the CPU. Most people would be more interested in the network switch performance than the CPU performance there though.
You are so much right. 1GBit has been dirt cheap for like 20 years. I remember when I got from 1MBit Ethernet to 1000MBit Ethernet in less than 15 years and after that: Nothing. Overpriced. I even used 40GBit Infiniband for a while to directly link my two most important systems and even that was a ton cheaper than 10GBit ethernet. This is utter desperation. My first computer to oversaturate a 1Gbit link was a cheapish Athlon 600Mhz with a cheap software RAID5 in 1999. My first computer to oversaturate a 10Gbit link was a lowest end Xeon with 2x2000Mhz and a single U2 drive in 2012. And now in 2021 we are still literally shitting bricks at 1Gbit.
I enjoy your reviews. I am data center focused myself, and always think about speeds, buffers, etc. For hosts that network boot, 2.5 Gig is potential interest. Typically we'll have 40 hosts 1gig, with 4x10 uplinks bonded (48 hosts if we want slightly over subscribed). With 40-48 hosts at 2.5Gig, I would need 6 25 gig uplinks (SFP) or 4 40gig uplinks. (QSFP). 10Gbase-T is getting cheaper all of the time. I can see hospitals moving over to this. If all of the nodes are 2.5Gig, we'll need better path to the boot/management nodes and upgraded nodes.
Personally, I use a switch to let me plug many devices into the router. I never use it for pc to pc file sharing, so 1 Gbps is more than enough for me. Half my connections are 100 Mbps, so I have no need for 2.5 X, but thanks for covering it.
Totally understand. There are a lot of 2.5GbE NAS units these days and 1.2Gbps down cable modems tend to have 2.5GbE links to help handle speeds over 1Gbps
I can reason with paying ~2X the price of a 1G switch. But no more. Frankly, I fully agree. If we are just about to hit the critical mass and in a few months to a year we have an influx of cheal 2,5G unmanaged switched, it would be a god send. If a was building a home network today, I wouldn't want anything less that 2,5G. 1G feels a bit limited for futureproofing.
CRS312-4C+8XG-RM from Microtik 99% fits the bill with a couple of well valued inclusions; 1. Managed 2. 8 Ports fully capable of 10.100/1000/2500/5000/10000 G all over copper and four SFP+ ports for the QFSPK 25G / 40G / 100G hobbist. Price per port for the over all value received is almost a 2:1 on this unit reviewwed here.... and no where near the feature set....
I don't have anything in my house that has 2.5GbE, so there is little need for me to go faster than 1Gb. About $120 for an 8 port 2.5GbE switch would be a reasonable price. At current prices I would be more likely to splurge on 10GbE gear and not have to worry about upgrading for a lot longer.
I just haven't had a need to move to 2.5Gb at this point. We use 10G for special high bandwidth points in the data center and 1Gb for everything else. If its really important it has multiple links for failover so technically most of our backbone is more than 1Gb anyway. You mentioned 2.5Gb on some newer Wireless APs, but I'd want POE for those, so unless as you say the price drops down to like 2-3 times that of existing Gigabit switches, then I might just go for it based on future compatibility, but for now its still too expensive.
These prices are reasonable imo, but I cant justify yet conidering my firewall is my only 2.5Gbit device, when I have at least two devices at this speed i might start looking at one
I agree, the jump is too high for "only" 2.5g. The price jump for a smaller 10g microtik is not much more so my thought would be to hold off until you absolutely need to jump up, then go directly to the 10g models.
2.5G 8x Port switch $89.99. 5G 8x port switch $169.99. I would prefer to see multi-speed ports though. I want a switch that will give me the fastest error free rate on the cable I am stuck with in my houses walls. Oh, if you add simple management, plus 25% price. Layer 3 management plus 50%.
For me, for my home network, I'd need two switches(Bedroom, Living Room), and if they were around $30 each I think I could pull the trigger, but I'd need 2.5g ports for 3, maybe 4 PCs, for those, maybe $10ish each(for USB/PCIe 1 port adapters)? My Router is currently only 1g, but I spent like $120 well over a year(maybe 2?) ago on a 'good' one(by home router standards) to last a while, so I'd prefer not to replace it. It has a USB3 port, not sure if it supports USB network adapters though. Not sure how I would organize things.
This is sort of drip feed.. We should be jumping straight from 1Gb to 10Gb really TBH.. We are way overdue if you look at the history of networking. ESXi I think doesn't even support 2.5/5Gbe
The thing is that most installed cabling right now wouldn't be able to handle 10G. NBase-T is a really great solution to stretch that infrastructure another 10-15 years. I'm a tech enthusiast , and honestly, gigabit is enough 99% of the time. BUT that is why I also have a few Cisco 9130s with 5Gbps links installed at home and wired to a POE++ switch. Should...should...I need it.
@@logikgr Yes, this is such an important aspect to those saying we need to just go to 10g or nothing. Consumers and buildings do not want to adapt to fiber/sfp's to get far runs connected at greater than 1g speeds. I think the price point of these switches would make more sense as a 5gbe switch, 5x bw increase beats the 3/per port to 23/per port metric posed by Patrick. This would also allow one link as a "backhaul" to another 5+ gb switch, and allow for a natural progression in the client device market from 2.5 to 5gbe eventually.
@@logikgr Ethernet can fall back to lower speeds if something is amiss in the chain. And a lot of cabling can do more than 1G, even if the cable isn't officially rated for it. A lot of people have done 10G over Cat 5e (ok, not at 100 meters lengths).
I think trendnet got their pricing right. It's a new consumer speed class with few (if any) devices in the home that require it. That makes demand low except for high end consumers and enthusiasts. Wifi 6 APs are in a higher price segment than these switches, so a price point like this on a matching switch is about right. But if we're being honest most consumers won't come in contact with either Wifi 6 or 2.5G until they get a new Wifi gateway from their ISP. Those volume units will drive down cost per port. In the mean time I think $170-$180 for an standalone 8 port 2.5G unmanaged switch in the consumer space is a fair price I'd happily pay if I were in the market for such a switch. I think asking for it to be cheaper is a normal reaction for any product, it would always be nice for things to cost less. But it's likewise unrealistic. Trendnet are making a good value proposition.
One of the major differences between the low end and the high end in terms of pricing structure is that with the high end, fiber infrastructure is abundant. Scaling vertically is as "easy" as swapping out your networking hardware. Your fiber is most likely already capable of handling the higher speeds. Additionally, at the high end we've always had more reasonable upgrade paths. You could go from 10G to 40G, and then eventually to 100G and 400G (25G/50G/200G were/are also options but they came a bit later). With the low end, it's basically a captured market. You want something better than 1GbE? Well, you're gonna need to upgrade your wiring (Cat6 has been around a long time now, but Cat5e is still ubiquitous). In most cases that's far more difficult than replacing fiber in a datacenter, and certainly adds a lot of cost to any potential 10G upgrade. Now that NBASE-T is coming to market we're no longer forced to upgrade wiring. But that doesn't change the fact that NBASE-T isn't really competing with gigabit. In fact, it's competing with 10GbE. The cost divide between gigabit and 10G is so wide, that there's plenty of room for margin in NBASE-T products. And that's reflected directly in the pricing (although low volume on new tech is surely a good portion of that pricing as well).
You can't assume the market's wrong here. I think the reason there aren't more cheap 2.5g switches available, is that nobody wants to go from 1 to 2.5 - it's not a big enough upgrade, even a partial upgrade. You're never gonna get mass adoption until the price of 10g hardware drops (even more).
can you please review the Zyxel Multi-Gig 12-Port? $150 for 2xsfp+ 10gb, 2x 2.5gbe , and 8x 1gbe. It seems like a great jack of all trades for the price, if using 2 of them as a backbone for a home network with nas in its own room, while most devices still have gigabit.
Glad to see progress, but consumer ethernet needs to stop screwing around. 5/7nm ICs mean we can get silent, low-power 10Gbe switches. PCIe 4.0 means we can get 10Gbe in a single lane. 2.5/5 should be there only when the line can't handle 10.
I couldn't agree more. 2.5/5 GB cann be a fallback if you have long old CAT5 cables. Otherwise it feel like a ripoff vs 10GB ports. I am surprised that after 15 year we still can't have mainstream 10GBE and Intel puts 2.5GB everywhere. Feels like a ripoff trend.
Regardless of how low power the switch ASIC itself uses, if you want 10GBASE-T you're going to be using a decent bit of power for it. The amount of power actually going over the cable is already 3W per port and even with some hypothetical 5nm ICs you're still not going to be able to manage anything close to that. 8 ports at 5W a port still leaves you with 40W of heat to get rid of, good luck doing that in a plain metal case without active cooling.
Unless you are buying a used Brocade switch or something, the price does not really compare, also 10GB over CAT6 is not power efficient at all so now your looking at SFP+ (which most people have no idea bout or where to start)
@@TheRealMrGuvernment "most people" arent even looking at 2.5G either. mikrotik has a 4-port sfp+ unit for less than $150, makes it a lot easier to step up to 10G, pcie sfp+ cards and twinax cables are cheap, so's fiber, because theres a premium for everything "Base-T" over 1G
Simple economy of scale... there's no market for it yet. Just like 10G, it was very expensive to start with. (if you were around eons ago, so was 100M. and the move from hubs to switches.) Give it time, the market will grow, costs will go down.
my nas is limited by the read/write of a single spinning rust drive so there is no point in upgrading my lan. If I upgraded my nas, I might consider a 2.5gbe but I honestly don’t have a need because I feel like the performance is very good right now.
I built a 8-HDD NAS and it was far from automatic (more RAM, flash cache) to get it to go beyond 2.5gb range, although it was pretty easy to go past 1gb.
A single disk of spinning rust can can transfer at around 180mb/s these days (and faster to/from cache) ... about 65mb/s faster than 1Gbe. I can say I was happy using 2X or even 3X 1Gbe connections to my NAS once SMB3 started giving me the combined bandwidth ... but even happier when I switched to 10Gbe and was able to free up some network ports.
as always, a great review. Even if the switch does not support any vlan mgt, you can ALWAYSs program your NIC in OS to support VLANs, with a VLAN ID. Just moving the management around a little :-)
@@Knightrider159 but the statement remains true either way.... If VLAN manip was available; that would put the switch in the managed platform It is VLAN aware and will move packets based on VLAN ID intra switch (VLAN1)
$10 USD per port is the maximum price for widespread adoption. It's just low enough to be rationalised by the average consumer. For enthusiasts on a budget, this is affordable and they'll not just buy it, they'll recommend it to all their friends. For enthusiasts with more money, 10GbE makes more sense.
I have been looking for a switch with the LEDs separate from the ports. I prefer it since then I can tape over them and don’t need to see the blinking. (For home use).
I went with a MicroTik SFP+ switch instead, crazy how expensive these are for the speed. I think $100-$120 is a reasonable price that would push mass market adoption. That's inexpensive enough that I could justify throwing that in at my folks house so the handful of devices with higher speed NICs they have could actually have fast access to the NAS I built them
Who are these switches intended for? My only guess is SOHO setups with a 2.5 gigabit NAS. Most consumers are fine with 1gig, and any larger businesses would most likely want 10gig uplink and/or management features. Compared to the Microtik CSS326 I don't see the value in this, but then again I don't have any 2.5gig devices.
I got a new motherboard with 2.5 and now have the itch to use it. There are cable modems with 2.5g now and some consumer routers with 10G wan. Too bad my wish list is nearing 900 bucks which is hard to justify for my casual use and occasional home lab fun
We need more competition or just go for 10 Gigabit and be done with it :)
Totally agree. I don't see the point for 2.5 and 5 Gig, if you still have to change your equipment
exactly
@@coso2 right, 2.5/5gb isn’t going to get me through 10 years (how I see the life of a switch). At this price point, if I need to buy something mid-way I might as well just go to 10Gb now.
Due to prices of 2.5GbE I've went with 10GbE hardware capable to work at 2.5GbE if required
@@coso2 99% of consumers barely use 1GB from there devices so internal networks for most home users or SOHO still don't need 2.5/5Gbps yet, but, companies need to make money so they will sell this as a stop gap and rake in the profits.
When you can get a whole 8 port gigabit switch for less than the cost of a single 2.5g port, you know something's wrong
There are cheaper options. QNAP makes a 5-port 2.5Gbe switch (the QSW-1105-5T) that has an MSRP of $100 USD (though real-world prices are more like $110).
That’s cheap enough. Back when, I was paying $2,000 for what was a 10mbe with seven of those ports with one 100mbe port. People complain too much.
@@guspaz It's cheaper, but still pricey. Street price for that in the UK is £110 (£22/port, about right for USD -> GBP + VAT); meanwhile I can get a TP-LINK LS108G for £16.
@@melgross I jumped into the GbE @ home world early on when 8 port switches needed a full U with active cooling and finally snuck under $1000. Just a few years later, they were half the size, 1/4 the power consumption, passive and $100. Now they're plastic and $25.
10GbT has stubbornly held its per-port price for damn near the last decade that I've been watching - over that same period 1g dropped 95%. Introducing 5g and 2.5g at "most-of" and "a-big-slab-of" the 10g port pricing doesn't make an especially convincing argument for.
@@mattd5136 while it may be hard to believe, Apple was credited for bringing the cost of both 100Mb and 1Gb down when they put ports into all of their machines. A year after they did that, cards dropped from about $750 to $100, and a year later, to below $50.
But Apple didn’t do that with either 2.5 or 10. They put 10 on their highest end, but lowest selling machines.
What’s needed is for a major computer manufacturer to make this standard. But no one has been interested. Unless this reaches major sales. Prices will just come down very slowly.
I'd love if motherboards, desktops, and other devices would start standardizing on 2.5 GbE as the baseline (whereas 1 GbE is now), but still, these switches are 4-5x more expensive than 1G units. Until that changes I'd still rather stick with hybrid gear like QNAP's 10/2.5 Gbps switch, so it's more future-proof for when I have more 10G-capable devices.
2.5 Gbps is very appealing for me since I already have 8 rooms in my house wired with Cat5e, and pulling/rewiring to Cat6a or Cat7 is going to be annoying.
It is like 7-9x more expensive than 1GbE which to me is the challenge.
Have you tested the existing cat5e runs with 10 GbE? If they aren't tens of meters long, it would likely work well. But of course it might be unreliable if the signal is stretched to its limits. You will definitely need cat6 to run the full 100 meters of 10 GbE. But at short distances anything that resembles an ethernet cable likely works just fine.
I agree with Jeff Geerling that the QNAP QSW-M2108-2C is the more attractive solution to me -- which is odd because just yesterday I was shopping for one on Amazon and reading a review about it there from a Jeff Geerling (and a week before that I was on Leanpub downloading Ansible books from a Jeff Geerling).
Anyway, I'd say 10 GbE uplinks on a 2.5 GbE switch are pretty critical here, as switches like the Mikrotik CRS309-1G-8S+IN make it pretty easy to have a (silent) home 10 GbE core nowadays, and a 10/2.5 GbE switch is a nice extension to that for the growing number of 2.5 GbE equipped gear.
I'm not sure if a standalone 2.5 GbE network has much of a value proposition to me, especially not at the price discrepancy you note. The question is: what are you connecting to at that speed? A 2.5 Gbps Internet uplink (I wish!), a local server file share (more probably), etc.?
I'd say that these days, working off a network file share at 1 GbE is just unworkable, and probably 2.5 GbE isn't enough of a step up. Local direct attached storage nowadays is commonly PCIe 3.0 NVMe, and increasingly PCIe 4.0 NVMe, so working over 1 GbE/2.5 GbE (essentially mechanical SATA drive speeds) just feels like your computer is freezing/broken. Compared to NVMe, even 10 GbE seem insufficient, clocking in at RAID 0 SATA 3 SSD speeds.
Really, we're approaching the point where 25 GbE is the lowest bar to keep the network competitive with local storage. What's more, 30m runs of OM4 fiber optics can be had for like $50 bucks. The justification for settling for slow copper network speed is getting slim. I'd sooner pay more for more PCIe lanes on a workstation (to have space to add 25 GbE+, RDMA capable NICs) and a silent home 25 GbE (even just 8 port) switch.
Does the QNAP QSW-M2108-2C support 2.5gb on the SFP+ ports?
@@ServeTheHomeVideo didn´t you see the 4x Intel lan cards from QNap that cost "near to nothing" .. make your own switch : QXG-2G4T-I225 (80€ here in germany)
Honestly, as a home user, I think 10USD per port will be my switching point. Currently, the only situation that saturates the current 1Gbps connection is bulk file transfers, which does not happen very often. I think what I care more about is consistent processing latency and stability.
would be nice if these switches had a 10gbe uplink.
I totally agree with this. I strongly prefer switches with faster uplinks.
Let's hope Mikrotik does something for this market soon
@@tommihommi1 A competitively priced Mikrotik switch with 8-24 copper 10GB ports (with 1/2.5/5 fallback), and two (Q)SPF+/28 25GB, 40GB or 100GB uplinks would be an instant buy from me. I don't even need POE.
@@kwinzman Why do you expect 2,5/5g fallback and 100gb from mikrotik so soon?
@@wiziek I do not. But I can articulate my preference and disappointment with the status quo.
$22 per port and they can't put the leds in the RJ-45 ports. The LED placement is an important issue. I totally agree with you on pricing. My guess is that we are just a year or two early for low prices. Thanks for the great videos.
i got the 5 port QNAp 2.5 switch was about 120 bucks and has leds on ports like normal
I thought the LED placement was nit-picking until I saw the price. These switches aren't going to age well as far as appearance. In one or two years they're going to look like the $25 bottom-of-the barrel switches we see today.
Agree that the per-port cost for 2.5g is quite a jump from 1g. It's high enough that I just went with 1g or forked out for 10g for the stuff that benefited from it for personal use. This would be perfect with an SPF+ port or 10g for uplink but without the £400 tag of the TEG-30102WS. Prices and product choice is slowly getting better but there's still some way to go yet.
$100 for 8 ports would be ideal (~$13/port), but $120 is probably the most I'd go. Let's be real, these will mostly be used as edge switches, not backbone switches. Since most residential internet connections in the US won't see even gigabit speeds for a significant period of time, and a gigabit switch with 10Gig uplink can be had for the current asking price of the 8 port unit ($189), I think they make very little sense at the current price point for home users like myself.
Unrelated; as much as I like to geek out about crazy high end equipment I can never own/have no use for, this kind of content (consumer/prosumer networking gear like unmanaged switches) is what I really need. Keep up the great work!
Glad they are finally here, however I feel like the benefit of going from 1gbe to 2.5g is too costly for enthusiasts in most cases. Meaning it makes more sense to go 10gig if going out and buying bespoke hardware. 2.5g needs to replace 1gbe as the default, with 10gig still the enthusiasts option I feel
I questioned that myself. I mean, we went from 10Mb Ethernet to 100Mb, and then from 100Mb to 1Gb, and then from 1Gb to 10Gb. Every jump was a 10x increase. But going from 1Gb to 2.5Gb... I doubt in most cases you could even tell the different. Like going from 100Mb to 200Mb. Not worth buying new equipment.
@@eosjoe565 What I don't understand is why 2.5GB pops up on all the new Intel motherboards? I mean you can buy USB 5GB NICs that are faster! Feels like penny pinching to me.
@@kwinzman You need to be careful with these 5GB USB NICs. A lot of them actually are not faster than 2.5GB. Seems to be a weird rip-off, but you should do your research before purchasing.
@@fbnx4219 I know that the qna-uc5g1t for example which links at 5GBE can only do 3.5Gbit/s because it's limited by the 5Gbit/s USB3.0.
But it acts as as a dirt cheap upgrade at 80 bucks, usable in pretty much any machine manufactured in the last few years, without even opening it. There is no risk of blue screening the computer when suddenly unplugging like with some Thunderbolt 3 NICs.
I suspect this problem will go away entirely with USB4. And then 2.5GBE on the motherboard will look even more outdated and peny-pinching cheap than it does already.
Enthusiasts got second hand fiber optical 10GbE gear years ago and don't have to care about this. I know I did.
Let these switch manufacturers milk the early adopters of this new market for all I care.
I mean, if you absolutely need the cable lengh of RJ45, these switches maybe make sense, but the 4x speed increase of the MikroTik 10Gbit switch with SFP+ ports(at the same price but with only four 10Gbit and one 1Gbit port) seem to make much more sense, if you can satisfy the bandwidth of course
At it's current pricing, per port cost is way too expensive for regular consumers who just want to plug and play. I agree with you on the recommended pricing.
I think you're right at the $10-12/port price point.
I think I would be buying some of these if they were a third of their actual price. Probably it will be 1-2 years from now, but I am quite confident that we will get there.
Mikrotik can make a 4x10GbE switch for U$S149, an 8x10GbE for U$S269 and the one that I bought the 16 port 10GbE with redundant power supplies for U$S399. And TRENDnet expects U$S180 for an unmanaged 8x2.5Gbe switch? I know, different customer base, but my point is... how big is the margin TRENDnet gets on these things? IMHO motherboard vendors were never interested in expending a single cent more than necessary because of the low margins they manage (not long ago having an Intel 1GbE instead of a Realtek based solution was the "high end" solution on expensive motherboards), and the lack on interest on the average customer never pushed them to do anything else. There are LOTS of marketing and money invested on WiFi technology, which before WiFi6 it was really "meh" technical wise, but the average customer wanted that... no cables, big numbers. It's a pity, really, the enthusiast has no choice but used enterprise NICs and this switch will be a market flop. It targets a market that... doesn't exists. An unmanaged 4/5 2.5GbE switch should be U$S50, and should be the norm. I updated my home network to 1Gbps... maybe 15 years ago. Sorry for the rant. It's a shame we are not using 10GbE as a cheap standard and replacement for 1GbE like 1GbE was to Fast Ethernet.
Still blows my mind that 15yrs ago I bought a netgear 8 port 1Gbps unmanaged switch for $75. Really wish I knew what the hold up is getting faster networking down to affordable prices. 2006 me was convinced by at least 2015 10GbE would be in the home. I was wrong.
Excellent points, Patrick!
I just spent the last few days doing a "drop-in" replacement of a Linksys 8-port Gigabit switch with an 8-port TP-LINK 2.5GbE switch. That new 2.5G switch caused zero problems. All the problems we encountered happened with 4 x USB 3.0 to 2.5GbE ethernet adapters, and 1 x 2.5GbE PCIe x1 add-in card -- trying 5 different vendors.
Because we still use Windows XP on our backup storage servers, we naturally had hoped the USB 3.0 dongles would work on those PCs. NOT!! Yes, I already hear you saying that XP is obsolete. HOWEVER, the Realtek website does list a download option for "WinXP Auto Installation Program" under the RTL8156 and RTL8156B controllers.
And, despite my good faith efforts to request Tech Support (now that I am a paying customer), what few responses I did receive from the sellers were terribly inadequate. The email address for Tech Support at asustor is "over quota" and my question to them is still waiting for answers.
I haven't told StarTech yet, but their dongle starts up at 100Mbps, and must be DISABLED and ENABLED in Network Connections to change the speed to 2.5Gbps. Also, that dongle required a hybrid USB cable that supplies extra power from a USB 2.0 port: the latter problem may be due to limited power available from the USB 3.0 ports at the real I/O panel of our HP Z220 tower workstation. So, a hybrid USB cable fixed that one fault.
The good news is that all 5 of those adapters installed mostly OK on Windows 7 and Windows 10 PCs. And, the excellent software at drivereasy.com made it relatively simple to upgrade device drivers to the latest versions.
If you're not already aware of the latter website, I have been updating drivers with their Windows program for several months now, and I have not had one single problem with any driver updates.
I should add that our new TP-LINK switch is on the floor, below one workstation monitor, and we point a DC fan directly at that switch because it does get warm to the touch.
Hope this helps.
p.s. Yes, Patrick, your astute analysis of prices is RIGHT ON THE MONEY: prices should naturally go downwards as awareness of 2.5GbE expands worldwide.
I should add that performance is what we expected, when other overhead sources are taken into account. For example, by watching the Ethernet port activity with Windows 10 Task Manager, the real-time graphs do approach a 2.5-to-1.0 ratio with averages of 10-to-4, more or less. This ratio is most obvious when we run XCOPY twice in a row, and file properties are mostly memory-resident during the second run of that Windows utility. Of course, if a Windows 10 PC uses a 2.5G controller to talk with a Windows 7 PC with a 1.0G controller, the overall speed is reduced by the latter "weak link" in that chain. What impressed me the most was the obvious speed jumps we saw when copying large driver software folders between 2.5G ports, particularly when both PCs were cabled directly to the TP-LINK switch, and did not need to traverse any switch "cascades" or pass traffic thru our router. With that experience under our belts, we're now planning to upgrade to TP-LINK model TL-NG421 PCIe x1 add-in cards, and keep the dongles in spare parts inventory, for now. We were delighted to discover that the latter add-in card also works in a PCIe Gen1 x1 expansion slot i.e. same raw bandwidth of 2.5G per x1 lane. Last thing: I'll be contacting our router vendor to ask if they have any plans to offer a drop-in 2.5G replacement for our 4-port Gigabit router.
This is great, although I wish it came with a one or two 10gbe ports.. It is still an "affordable" switch for what it offers. Its nice to see someone taking the time to review network gear -- This is exactly why I come to this channel -- Thank you, Patrick! :)
I think the idea here is that if you've got that much traffic to pump a10G port, you'd have the coin to upgrade to a more powerful switch overall
You get a like for showing the internals!
That is what we do!
"We did a rocket-lake review that nobody watched"
*opens the rocket lake review and watches it, that's how I roll
Thanks!
For me personally, I'd love to upgrade my small homelab to something faster than 1Gb, but for it to be really feasible for me, I think things would have to hit the $10-12 per port range or
For a home what I am looking for is 18 ports. 4 x 2.5gbe poe for wifi, 4 x 1gbe poe for cameras, 4 x 2.5gbe for desktops and nuc, 4x 1gbe for tvs and other legacy devices, and 2 x 10gbe for servers/ Nas.
Maybe a 26 port that you double the nonPOE ports.
I think $10/port gives them enough of a profit over the $3/port for current unmanaged switches and still feels like a fair price as a user. Especially since OEM's are now including 2.5Gb Ethernet as standard equipment. It's no longer a special order item like a 10/40GB device.
The TEG-S380 seems impressive, but there's been a cheaper 5 port option for around a year now, the QNAP QSW-1105-5T, which is $110.
The QNAP always seems to be on backorder....you know of any sites that has 1 in stock for $110?
@@thecockflock You can always put an order in at B&H and just wait for the backorder to come in.
@@guspaz The feasibility of doing that would be related to how urgently they would need the switch.
@@thecockflock At the time of writing this, All Qnap has some, but for $124 instead of $110. My point is that it's still cheaper than the $140 for the TEG-S350.
@@sagetechnology4913 Sadly still way above the goal of $10-12 per port that they were talking about
After looking for years for faster than 1GbE solutions, I finally made a jump to 10GbE over fibre. Cards are super cheap on eBay and 2 cheap Chinese SFP+ modules + fibre patch cable is cheaper than a DAC cable or a pair of -T modules.
You see it the same way that I do ... 11:38 ... "That gap is too much". I agree. That's why I don't think that 2.5Gb will take off. Most people rather spend a little more and get 10Gb. At the price point of $24/port, that is just stupid and ridiculous. Way too much to make it universally acceptable in its pricing right now. Basically, 10G is gonna be accepted faster due to these high prices.
If you look at the history of consumer and some business hardware, (here in the UK), you will see that even in the first decade of this century motherboards, until later in the decade were still at 100 mbps, though Laptops/notebooks had gone up to a Gig. That meant a lot of places still had 100 MB infrastructure with 1 Gig uplinks. It was only as more items got 1 gig that the market moved. So before the end of that decades where I worked we had gone to 1 Gig throughout with a couple of 10 Gig links, but hey were expensive. You found alternative way to increase uplinks, etc bandwidth if required. These days I only run a small home lab and that is still on 1 Gig and is OK for my current requirements though will look at going higher when the prices significantly reduce, or my needs change. The benefit of 2.5 gbps is that I can continue most likely to use my CAT5E cabling, and I have boxes of that in the shed, even maybe be able to handle 5 gig at a pinch. However what would be interesting is due to reduced costs of production as sales increase and prices come down will there then be that much difference in costs between a 2.5 and 5 gig switch on a per port basis.
Incidentally all my switches are metal cased. The only item that isn't is my router that connects via copper to FTTC. Regarding the port indicating LEDs. I do agree that above the port is best. Where I worked we had racks of switches and most indicators were above the port and they were so easy to detect that they were fully operational, and at what bandwidth , than the few that were to the side of the ports.
For my home network I would struggle to spend much more on a 2.5gb switch than 1gb. The way I'm thinking about it is the use case for every device needing >1gb would be if I have an internet connection that can saturate it, but I don't so realistically there are only a few devices passing around large amounts of local traffic. If the amount of traffic is noticeably slow on a 1gb connection then at that point I'm just going to move those devices to 10gb.
For an 2.5GbE 8 port unmanaged switch around 100USD is what I would be willing to spend. Around the 200USD mark I would rather just hop on ebay or similar sites and pick up a used managed switch. But to be completely honest I wish companies would just skip 2.5GbE and go right to 5 or even 10GbE.
Intel pushes 2.5GB on all their new motherboards, I don't understand it either. Must be saving a few cents vs 5/10GBE.
I agree with you, 5GbE is around the same speed of SATA SSD's (~550Mb/s) and also USB 3.0 speed. Should be bare minimum for 2021.
2.5Gbe is the most cat5e can do.
@@kwinzman i don't mind intel doing this as much, as we can add network cards. Would be better if they went to 5GBE, but at least they are leaving the GB phase. However, we hardly have any option for switches at 5GBE (or more) that does not cost an arm and a leg...
I would have an easier time paying more overall for a cheaper 5GBE switch + adding cards on each machine (but not all at the same time, perhaps 1 or 2 each month) than having 5GBe on all machines from the start, but paying a high price upfront for the switch...
Obviously would be better if 5Gbe or more were the intel move AND switches were cheap :)
@@AudreyRobinel And even rarer is to see an actual 25GBE SFP28 uplink or ideally 2 SFP28 uplinks on your 5/10GBE access switch. For something that would actually make sense on the switch side you suddenly pay 100 times more than for similar 1GBE switches with SFP+ uplinks.
Heck. Mikrotik's 10G offerings (4+1, and 8+1) at $140 and $256 respectively... managed, fanless... very competitive as well, as long as you're buying DAC cables and not having to invest into SFP modules. It's good to see new 2.5G offerings, but I agree the price needs to come down further.
you can find used QSFP switches around that price point too if you want to go for 40gb...
@asdrubale bisanzio indeed
@@callowaysutton those are all super noisy… unless noise isn’t a concern.
Plus the Mikrotik switch allows to have dual PSU. That's crazy good at this price.
Your pricing (16:50) is somewhat right. They should charge double say what 1Gb switches are, say $55 for FIVE 2.5Gb ports and
then (maybe) $85 / $90 for EIGHT 2.5Gb ports. If not, I'd buy 10G switch instead. It has to be HALF what they are charging now.
I recently purchased a Ubiquiti Enterprise 24 PoE for my home with 12x1Gb, 12x2.5Gb, 2x10Gb SFP+ ports, at $800. I love it because it supports Comcast's 1.2Gb internet at full speed, and has PoE for security cameras and APs. Not all of the ports are 2.5Gb, but for the security cameras and non-core equipment that's not a concern. It seemed like a good compromise, especially for the PoE.
I think if you go over $100 it has to have compelling features such as a 10Gb uplink or be managed. An unmanaged 8 port 2.5 with a single 10 gig (sfp+) at around 150 would be right around where I would be ok with it. I would prefer it nicely in the 120-140 range. A 5 port unmanaged 2.5gb switch needs to be under $75 for general adoption or it needs to have a higher uplink to be any more expensive.
Honestly, at this price it’s still too much. For $140 I’d rather do the tiny mikrotik with SFP+. $23/port is about 3 times as much as what it’s worth.
You mentioned that a gig port is $3 so makes sense that a 2.5G port should go around 2.5x$3=$7.5/port.
Anything above $10 a port makes no sense.
I would happily pay $60 for an 8-port 2.5G switch. That would be awesome.
Im pretty much at the same page as you are talking about, i want a 2.5gb switch, but they are far to expensive atm to make it an option, versus just sticking to a 1gbit switch, it is faster sure, but for my use, it wouldnt make a hell of difference, at least to a point where a 10x price premium is worth it. To be relevant pricewise it would have to come down to something like $60-$70 for a 5 port switch.
I don’t want to spend over $100 for a 4/5 port unmanaged switch. At that size its just going to be home use plugged in and forgotten about. That comes out to 1 uplink and $25 per device added to the wired network.
With 1G switches at $2 - $3 per port, going to 10x the price for 2.5G is totally unreasonable. Yes, they’ll definitely sell some at those prices, but sales volumes won’t go up until they get it down to around $10 per port. I’m pretty sure for most consumers, 2.5 times the performance is just not worth much more than 2 or 3 times the price.
I agree to 100% , and this as someone who do e the change to 10gbit 2 years ago. That might be different for 5gbit, as even with 10gbit you have to decide if you rather do the step to SFP+ anyway. The switch manufacturer just missed the opportunity to really make a step to 5gbit and offer a real compromise for consumer. With the same cabling able to handle the 5x throughput.
Edit: Also, I absolutely won't be buying any of these until I can get a 5-port switch for $60 max. I think your $12/port assessment makes perfect sense.
I completely agree with your thoughts on the price jump from 1Gbe to 2.5Gbe. I don't really see the justification...it doesn't seem like the parts or engineering work are dramatically different from 1Gbe. So ... it seems like the prices make no sense. It should be *way* cheaper. Like less than half lol
Even though it's unmanaged there are a lot of unmanaged switches that supports vlan. Would be good to know if by any chance, it supports for VLANs.
Unmanaged means one does not have a management interface on the switch to set VLANs there. One can still set VLANs on other switches/ NICs in the infrastructure.
If by "supports vlan" you mean passes the vlan tag through unmodified then all dumb layer 2 switches "support vlan". The switches that don't are using a switch chip that can handle tagging and untagging vlans but it's just configured improperly and being unmanaged you can change it from how the manufacturer configured it. Likewise some switch chips out there support STP but can be configured improperly to effectively just eat BPDU messages and forward all traffic thus breaking STP for those ports when hooked up to a switch that supports STP. As far as a dumb layer 2 switch is concerned, 802.1q frames are valid regular ethernet frames. The source and destination mac addresses are in the right spot, the first 16 bits of the 802.1q tag are 0x8100 so as far as the switch is concerned, this is the ethertype field to pass along unmodified. It extends the max length of a frame, but modern switch chips generally all support larger frames anyways.
@@ServeTheHomeVideo Just a few weeks ago I read about a switch that was unmanaged but had a dip switch that when enabled put each port in a separate VLAN. A switch like that with a 10G uplink (and preferably 2.5G ports) would be awesome.
@@ServeTheHomeVideo sure, it's that some tp-link switches has a "management" ip that you can access and do Vlan tagging, but they are still considered unmanaged. Tp-link used to call them Smart Switch... Idk if I'm wrong but for SOHO just being able to do Vlan tagging is great!
@@AndrewMerts I expressed it poorly. What I meant was that I've seen some SMB switches that are not managed but can still do Vlan tagging. I used to have a SG108e from tp-link that could do Vlan tagging, was not a managed switch, and it couldn't do LACP for example.
My current problem with 2.5GBE is that while it is faster than GBE, it is not super fast. It is aproximately good hard drive fast, or a bit faster. But it is way slower than even sata ssd speeds, and let's not even talk about nvme. So going 2.5GBE for me is already a compromise, i settle for lower speed that what i would want. I could be ok with that if it were cheap, but it is still quite expensive... So i feel like i am accepting "MEH" performances and still paying premium...
I would feel more satisfied by paying twice as much, but for 10GB lan.
The QSW-308S is an 8-port gigabit switch with three 10 gigabit SFP+ ports for $139 USD. They have an excessive number of variants of it based on if it's managed or not, or how many SFP+/10 gig copper ports it has. Add management and a fourth SFP+ port for $40, or add a secondary 2.5/5/10 gig RJ45 port to an SFP+ for $30 a pop (it's cheaper than a copper 10 gig SFP+ module). They range from $139 for that cheapest unmanaged one (QSW-308S), to $299 for the managed one with four SFP+ and four 10 gig copper ports (QSW-M408-4C) or the one with eight 2.5 gig ports and two SFP+/copper 10 gig ports (QSW-M2108-2C). Microtik also has a $99 switch that's an 8-port GigE with two SFP+ ports.
@@guspaz Hello! this is a really interesting reference!
I currently own the mikrotik 99$ switch (and in retrospect, i should have bought the larger version, with 24 gb lan and still 2 SFP+ for 140$).
The QNAP offer is really interesting, for now 3SFP+ would totally do it for me... and with four i'd be set for my next phase (a TrueNAS box, while keeping the previous OMV nas).
300$ is a bit more that what i'd like to pay, however, for 8 10Gb ports, copper and SFP+, this is really nice. If i'm in at 200$ perhaps adding 50% more and getting full 10Gb is the way to go. Plus copper 10Gb means that i don't have to buy cages. However 10GbE cards seems more expensive than 10Gb SFP+ cards.
@@AudreyRobinel If you already have the $99 microtik one, you may want to simply buy their 4xSFP+ switch ($150, CRS305-1G-4S+IN) and connect it to your existing switch, either with a 1 gig uplink via the 305's gigabit copper port, or via one of the SFP+ ports with a direct attach cable. However, that would only get you two additional SFP+ ports over what you have now, due to the two ports consumed by the uplink connection. The $270 CRS309 solves that problem by having 8xSFP+ ports, but now the cost is starting to get out of control.
@@guspaz that was the plan!
with both, i'd have a total of 4 available ports (2+4 -2 for uplink). But i have been looking at the 8 ports variant too, since if i'm in for 150$, i may well add 120 more and be set for a long time (i can see how i fill 4 SFP+ ports, but i don't have plans that would fill 8 for quite some time. So nice headroom for my use case here!)
I have also taken note of their 24 SFP+ switch, with 2 40Gb links :)
Way too expensive for what i am doing, but at 500$, this seems like an amazing per port value, and 40Gb links is something i didn't even imagine that would ever be in my price range.
the 40Gb link would be nice for a SSD cache on the NAS :D (but then i need to add a pool of fast NVME drives, a 40Gb lan card to the NAS, probably a beefier CPU, more ram, etc... gets out of hand quickly ^^)
Same thing happened when the transition on the desktop from 10/100 to 1000Base-T occurred. Will probably be another year or two and we'll see 2.5G at less than $100; nice that they're fanless. So many of the first cheap gig-E switches that would fail from overheating due to crappy or clogged fans. Back then it was a little different in that a home NAS was not anywhere near as common. The upshot of this is that these switches with no higher-speed uplink ports are really only for the smallest and most basic networks while the one's like you mention with dual sfp+ ports now have a much bigger market. A managed switch with fiber capable 10G ports for a ~$250 is a steal. Make the jump people!
20 years ago I wired a house for 10/100. That infrastructure still works for 1 GbE. I might be able to get 1 Gbit Internet here, but I don't feel all that constrained by the ~100 Mbit I usually get.
My currently router is limited to 1 GbE. I can bond ports on the back side to get a bit more bandwidth between the house network and my home lab, but for that use 1 GbE is good enough. I will eventually need to find a way to get 2.5 GbE to the router from the WLAN, but not until I upgrade the APs.
The key problem for 2.5 GbE for me is that it really only makes sense if you are regularly saturating 1 GbE, and if that case, you are probably going to want 10 GbE (or better) anyway.
I agree that the $100-120 range is the point at which these go from being "early-adopter" to "worth considering." It'd still be great to see them get below 3 figures, though, then things like that QNAP with the 10GbE uplink ports won't seem so attractively priced.
Motherboards that include 2.5GbE still tend to be on the higher end of the scale, roughly $150 and up, so there's still some of that early adopter premium in play. It's still good to see the new NUCs adopting it.
My B550 board has 1x1GbE & 1x2.5GbE, so I've been tempted to try it out. Plan A was to get an 8-port 2.5GbE switch, for future expension, and add a PCIe card to my Home Lab rig (currently just one system with a far-too-loud PSU). After spending ages looking for an attractive 2.5GbE switch, I gave up and ended up buying a Netgear GS108Ev3 for about $30 to play with the switch management features (I've only ever had unmanaged switches). Plan B is to stick with 1GbE for the rest of the network and have a direct 2.5GbE link between the home lab server and main PC. Still, finding a 2.5GbE card that's reasonably priced (and doesn't look like it'll break if you look at it funny) is proving difficult.
I really appreciate that you covered power consumption since 10gig switches are power hogs compared to 1gig.
Most home users won't need 2.5gbps and the power users have already gone to 10gbps. 2.5/5gbps equipment is DOA. Most systems are coming with dual 1gbps ports and you can bond ports and be way more cost effective. You can get dual port USB3 adapters for cheap.
I agree. I wouldn't consider buying into 2.5G until I can find a 8 port below 100$. It's not like 2.5G is for enthusiast when we even have ports on NUC systems, and for mainstream adoption the 100$ barrier is kind of a big deal.
Gees port costs are so high,
The switch prices need to come down to $10 or less per port
I remember when 2.88 MB floppy disks came out but then disappeared without widespread adoption because folks were not willing to pay double. I'd pay $50 for a 5-port switch, but I'd have little interest in upgrading whilst my internet connection speed is capped at 200 Mbps.
I stumbled over (didn't try yet but think about getting) the Zyxel XGS1010-12 which I think is another interesting unmanaged switch - same price range but it enables interconnecting 1gbit, 2,5gbit and 10gbit sfp+
Love the video. I really am holding out for 10Gb with support for 5/2.5Gb instead. I want to have the higher speed uplink to the rest of my network
10 ports please: 8 x 2.5GbE (4 w/ PoE), 2 x SFP+. $200. It's OK I'll wait.
QNAP QSW-M2108-2S for $250, that is pretty close.
Not sure if I could go fully unmanaged so bought the zyxel xgs1210 so I could get 10 g uplinks with couple of 2.5 gig ports, with vlan support for about the same price as these.
This was a fantastic review however it should be around 10 per port this is overpriced. So for me to upgrade I really want a 16 port managed version of this switch for around 160 to 180 max. So in short I see zero value at the current pricing IMO you should just purchase a 10 gig switch instead of wasting money on an overpriced 2.5 switch.
I'm doing switch upgrades for our whole small office network this year. The killer product for me would be a 48-port layer-2 1u multigig switch with SFP+ (at least) uplinks and no PoE for about $1000. It somehow seems ridiculous to me that I'm pretty much stuck with 1g access ports in 2021, without spending at least $4k/switch, where I can get the equivalent 1g switch for about $500. And the office isn't quite small enough to run off just one 48-port model...
You can get two CRS326-24S+2Q for $1000, probably not what you need since it has no RJ45 ports but you get two switches with 24SFP+ 10G and 2 QSP+ 40G
@@bjornSE That is a good deal, but doesn't work as an access switch - at least not without $40/port worth of SFP+ modules. Seen some good deals out of Mikrotik, but I've also heard of a decent rate of software problems on their switches, so not something I'd bet my reputation on quite yet.
The market for 2.5 Gb is very limited in the consumer market. Wifi has improved significantly and for the wire folks, the logical upgrade path from 1 Gb is 10 Gb.
Would love to see 8 port 2.5Gbe switches with 2 10Gbe ports coming in under $200, that's auto-buy territory for me!
I think I'd not buy a 2.5G switch. Either I can use the speed, then I'd get a 10G / 25G / 100G switch or speed isn't critical and there I'll just stick to 1G which is still plenty for Internet connectivity or accessing local services on the network from a desktop. But maybe other people actually do have use-cases for it.
Sound reasoning, for those with a need for quick access to large amounts of data 10Gb should have been a no-brainer years ago. Or we're in the Facebook/UA-cam/Netflix crowd and for those 1Gb is more than enough with even a very high-bitrate stream from one of the mentioned services less than 100 Mb/s.
Personally, I think 2.5Gbps is a waste of everyone's time and money. Just go straight to 10Gbps. The fact that even SATA SSDs are bottlenecked by it makes it feel old on arrival.
It feels like a complete ripoff. I am surprised how hard Intel pushes 2.5G vs 10G
The downside of 10G (at least as it is) is its massive power draw. And in some countries that makes the difference.
No-one has created at 10GbE copper device that works on Cat6 @ 100m, I don't think? People don't want to change all the cables in walls, that's a big driver for 2.5GbE. Along the same lines: WAPs that push > 1Gbps and use PoE.
@@SuperSpecies Cat 6A supports 100m 10GBE and it's not that expensive. If you have legacy cables that are really pushing the limit on the length you can always auto-negotiate NBASE-T on the 10GBE NIC. 2.5GB is penny pinching. Give my clients NICs that support up to 10GBE by default.
@@kwinzman many installations have cat 5e and cat 6 installed, and replacing the wiring in the wall with cat 6a is a massive capital expenditure and logistical nightmare at some places. Sure a new site go with cat 6a or better, no doubt.
I went straight to 10G on used hardware since it was cheaper than buying new 2.5G. If it's gonna take off I feel like 2.5G needs to beet the used market
It will never do . Its there for peopl that have no clue at all. You also find 10G overpriced consumer stuff. 2.5G existed for companys that didnt want to replace there old cabels but still go a bit faster . 10G is a sweat spot in my opinion
Maybe for 802.3bz 5 gig ports, but really the standard for Home /SMB should be ten gig to really justify the expense of upgrading the capacity of an entire network. Right now I use 1 Gig due to price and the fact that my home network uses L3 switches as I run IS-IS routing protocol and dual stack IPV4/IPV6 for all my network devices. I seriously doubt I am a typical use case though.
We should be pushing to 10G and not wasting time on a stop-gap 2.5g. Protocol that won't really live long enough to make it into mainstream devices.
It's like a 250mbps port instead of a1G
I'm currently looking at an upgrade project were I would like to add a 2x port 2.5Gbps Ethernet NIC to a small group of workstations and bond the channels for a maximum transfer rate of 5Gbps. If the price per port was in the 12 - 13 dollar range then the cost of two switches and the NICs would be acceptable.
The price point of 2.3Gb Switches is part of why I didn't worry about getting 2.5Gb Ethernet on my motherboard when I built this machine a year ago. Switch prices are just too high to justify using it yet for me.
Sure, in a fantasy world they wouldn't cost anymore than 2.5x the 1Gbps, but I also realize from the manufacturers perspective there are economic and efficiency reasons that may not be entirely feasible. I also tend to go out of my way and spend a little more on 1Gbps un-managed switches to get things like a metal frame, maybe some additional mounting options, and possibly a little more reliability in the hardware given the price.
Before I heard you throw out your number, I was thinking around 4x what the cheap 1Gbps switches normally cost (with a little less margin for the type of switch I usually go for), which ended up being in the same ballpark as yours.
Waiting for a 20 port with 4xSFP+10 for ~$300 to replace my trendnet 20+4SFP+ that i bought several years ago for ~$250
I too dont like the seperate pannel for LED activity/link state
48, 24 ports switches around my place.
Only use 5 port switches for a week when testing temporary equipment
To start with paying double the 1GbE equivalent is the pass fail line for most of my casual use acquaintances, but over time that has to come down. For the more advanced 2.5GbE isn't worth the hassle and they're looking to 10GbE, even if they have to go used.
I would think of that high cost as a so-called "early buyers tax", since I still find 2.5GbE a novelty. But they will definitely come down in price, just when it will would be a different and relatively unanswerable question at this point.
Where it makes sense now is if you have a home network storage, where the higher network connectivity bandwidth means faster data transfers. But once the Internet side of things transition to 2.5GbE like modems, optical network terminals and routers, the prices might go down further as manufacturers push out more products and more consumers gain access to higher speeds.
Yes, exactly. As I have commented before this unmanged switch without faster uplinks worth maximum $20.
Maybe 10-20% more than an equivalent 1G switch. Until price becomes resonable it is just milking uninformed consumers that don't know about 10GBE.
@@kwinzman yep, many would be better off buying 10GbE, would probably cost the same and yet future-proofing.
My current ISP even offers 10Gbps plans over fibre for a cool $170 SGD. But only a few people would be able to saturate the connectivity, since even I could run on about 300Mbps just fine as I am stuck on using powerline.
@@kancheongspidergaming Interresting perspective! Personally my use case is NAS. Even a cheap SATA SSD is already too much for 2.5G Ethernet, 10G is the logical upgrade from 1GBE. But 2.5G is getting pushed by big Intel now so it might survive if it's the same price as old 1G.
For switches, do people do benchmark on it? Like some kind of "query per second" kind of benchmark?
On unmanaged switches we can basically do throughput, we could do pps. On higher-end switches, one tends to have a CPU, memory, and storage running Linux so perhaps one could run a database benchmark on the CPU. Most people would be more interested in the network switch performance than the CPU performance there though.
@@ServeTheHomeVideo Ahh ok , cool thank you so much for the info!
You are so much right. 1GBit has been dirt cheap for like 20 years. I remember when I got from 1MBit Ethernet to 1000MBit Ethernet in less than 15 years and after that: Nothing. Overpriced. I even used 40GBit Infiniband for a while to directly link my two most important systems and even that was a ton cheaper than 10GBit ethernet. This is utter desperation.
My first computer to oversaturate a 1Gbit link was a cheapish Athlon 600Mhz with a cheap software RAID5 in 1999.
My first computer to oversaturate a 10Gbit link was a lowest end Xeon with 2x2000Mhz and a single U2 drive in 2012.
And now in 2021 we are still literally shitting bricks at 1Gbit.
I just pulled the trigger on the 5 port switch for 130 bucks I feel a little bad but I really like the speed to my unraid server
I enjoy your reviews. I am data center focused myself, and always think about speeds, buffers, etc.
For hosts that network boot, 2.5 Gig is potential interest. Typically we'll have 40 hosts 1gig, with 4x10 uplinks bonded (48 hosts if we want slightly over subscribed).
With 40-48 hosts at 2.5Gig, I would need 6 25 gig uplinks (SFP) or 4 40gig uplinks. (QSFP).
10Gbase-T is getting cheaper all of the time. I can see hospitals moving over to this. If all of the nodes are 2.5Gig, we'll need better path to the boot/management nodes and upgraded nodes.
Personally, I use a switch to let me plug many devices into the router. I never use it for pc to pc file sharing, so 1 Gbps is more than enough for me. Half my connections are 100 Mbps, so I have no need for 2.5 X, but thanks for covering it.
Totally understand. There are a lot of 2.5GbE NAS units these days and 1.2Gbps down cable modems tend to have 2.5GbE links to help handle speeds over 1Gbps
I can reason with paying ~2X the price of a 1G switch. But no more. Frankly, I fully agree. If we are just about to hit the critical mass and in a few months to a year we have an influx of cheal 2,5G unmanaged switched, it would be a god send. If a was building a home network today, I wouldn't want anything less that 2,5G. 1G feels a bit limited for futureproofing.
I hope mikrotik brings a 4-8 port 2,5g with 1-2 sfp+ for around 150... This would be perfect
CRS312-4C+8XG-RM from Microtik 99% fits the bill with a couple of well valued inclusions; 1. Managed 2. 8 Ports fully capable of 10.100/1000/2500/5000/10000 G all over copper and four SFP+ ports for the QFSPK 25G / 40G / 100G hobbist. Price per port for the over all value received is almost a 2:1 on this unit reviewwed here.... and no where near the feature set....
@@tmcarter3 and it is really cheap....
@@DarioSeibold well... as with most things.. cheap is not always best :-)
I don't have anything in my house that has 2.5GbE, so there is little need for me to go faster than 1Gb. About $120 for an 8 port 2.5GbE switch would be a reasonable price. At current prices I would be more likely to splurge on 10GbE gear and not have to worry about upgrading for a lot longer.
For those looking for 10Gbe, check out this 8-port switch: *Netgear XS708E v2* You can get them used in the $400 range.
Looks great, but it needs two 25GB/40GB or 100GB uplinks. I don't like SFP+ on 10GB access switches.
I just haven't had a need to move to 2.5Gb at this point. We use 10G for special high bandwidth points in the data center and 1Gb for everything else. If its really important it has multiple links for failover so technically most of our backbone is more than 1Gb anyway. You mentioned 2.5Gb on some newer Wireless APs, but I'd want POE for those, so unless as you say the price drops down to like 2-3 times that of existing Gigabit switches, then I might just go for it based on future compatibility, but for now its still too expensive.
These prices are reasonable imo, but I cant justify yet conidering my firewall is my only 2.5Gbit device, when I have at least two devices at this speed i might start looking at one
I agree, the jump is too high for "only" 2.5g. The price jump for a smaller 10g microtik is not much more so my thought would be to hold off until you absolutely need to jump up, then go directly to the 10g models.
It’s interesting but when options for 5 or 10 Gb switches are not to much more it would make the 2.5 switches irrelevant at those prices
2.5G 8x Port switch $89.99. 5G 8x port switch $169.99. I would prefer to see multi-speed ports though. I want a switch that will give me the fastest error free rate on the cable I am stuck with in my houses walls. Oh, if you add simple management, plus 25% price. Layer 3 management plus 50%.
Perfect feedback.
For me, for my home network, I'd need two switches(Bedroom, Living Room), and if they were around $30 each I think I could pull the trigger, but I'd need 2.5g ports for 3, maybe 4 PCs, for those, maybe $10ish each(for USB/PCIe 1 port adapters)? My Router is currently only 1g, but I spent like $120 well over a year(maybe 2?) ago on a 'good' one(by home router standards) to last a while, so I'd prefer not to replace it. It has a USB3 port, not sure if it supports USB network adapters though. Not sure how I would organize things.
This is sort of drip feed.. We should be jumping straight from 1Gb to 10Gb really TBH.. We are way overdue if you look at the history of networking. ESXi I think doesn't even support 2.5/5Gbe
The thing is that most installed cabling right now wouldn't be able to handle 10G. NBase-T is a really great solution to stretch that infrastructure another 10-15 years.
I'm a tech enthusiast , and honestly, gigabit is enough 99% of the time. BUT that is why I also have a few Cisco 9130s with 5Gbps links installed at home and wired to a POE++ switch. Should...should...I need it.
@@logikgr Yes, this is such an important aspect to those saying we need to just go to 10g or nothing. Consumers and buildings do not want to adapt to fiber/sfp's to get far runs connected at greater than 1g speeds.
I think the price point of these switches would make more sense as a 5gbe switch, 5x bw increase beats the 3/per port to 23/per port metric posed by Patrick. This would also allow one link as a "backhaul" to another 5+ gb switch, and allow for a natural progression in the client device market from 2.5 to 5gbe eventually.
@@logikgr Ethernet can fall back to lower speeds if something is amiss in the chain. And a lot of cabling can do more than 1G, even if the cable isn't officially rated for it. A lot of people have done 10G over Cat 5e (ok, not at 100 meters lengths).
I think trendnet got their pricing right.
It's a new consumer speed class with few (if any) devices in the home that require it. That makes demand low except for high end consumers and enthusiasts.
Wifi 6 APs are in a higher price segment than these switches, so a price point like this on a matching switch is about right.
But if we're being honest most consumers won't come in contact with either Wifi 6 or 2.5G until they get a new Wifi gateway from their ISP. Those volume units will drive down cost per port.
In the mean time I think $170-$180 for an standalone 8 port 2.5G unmanaged switch in the consumer space is a fair price I'd happily pay if I were in the market for such a switch. I think asking for it to be cheaper is a normal reaction for any product, it would always be nice for things to cost less. But it's likewise unrealistic. Trendnet are making a good value proposition.
About $8 to $10 per port for 2.5 sounds good. Personally, I'm waiting for 10Gb switches to come down a little.
One of the major differences between the low end and the high end in terms of pricing structure is that with the high end, fiber infrastructure is abundant. Scaling vertically is as "easy" as swapping out your networking hardware. Your fiber is most likely already capable of handling the higher speeds. Additionally, at the high end we've always had more reasonable upgrade paths. You could go from 10G to 40G, and then eventually to 100G and 400G (25G/50G/200G were/are also options but they came a bit later).
With the low end, it's basically a captured market. You want something better than 1GbE? Well, you're gonna need to upgrade your wiring (Cat6 has been around a long time now, but Cat5e is still ubiquitous). In most cases that's far more difficult than replacing fiber in a datacenter, and certainly adds a lot of cost to any potential 10G upgrade.
Now that NBASE-T is coming to market we're no longer forced to upgrade wiring. But that doesn't change the fact that NBASE-T isn't really competing with gigabit. In fact, it's competing with 10GbE.
The cost divide between gigabit and 10G is so wide, that there's plenty of room for margin in NBASE-T products. And that's reflected directly in the pricing (although low volume on new tech is surely a good portion of that pricing as well).
You can't assume the market's wrong here. I think the reason there aren't more cheap 2.5g switches available, is that nobody wants to go from 1 to 2.5 - it's not a big enough upgrade, even a partial upgrade. You're never gonna get mass adoption until the price of 10g hardware drops (even more).
can you please review the Zyxel Multi-Gig 12-Port? $150 for 2xsfp+ 10gb, 2x 2.5gbe , and 8x 1gbe. It seems like a great jack of all trades for the price, if using 2 of them as a backbone for a home network with nas in its own room, while most devices still have gigabit.
Glad to see progress, but consumer ethernet needs to stop screwing around. 5/7nm ICs mean we can get silent, low-power 10Gbe switches. PCIe 4.0 means we can get 10Gbe in a single lane. 2.5/5 should be there only when the line can't handle 10.
I couldn't agree more. 2.5/5 GB cann be a fallback if you have long old CAT5 cables. Otherwise it feel like a ripoff vs 10GB ports. I am surprised that after 15 year we still can't have mainstream 10GBE and Intel puts 2.5GB everywhere. Feels like a ripoff trend.
Regardless of how low power the switch ASIC itself uses, if you want 10GBASE-T you're going to be using a decent bit of power for it. The amount of power actually going over the cable is already 3W per port and even with some hypothetical 5nm ICs you're still not going to be able to manage anything close to that. 8 ports at 5W a port still leaves you with 40W of heat to get rid of, good luck doing that in a plain metal case without active cooling.
Too expensive for the common folks, too slow for the nerd where price isn't as important. imho.
Sorry, at this price point, I prefer to go 10 GB and be done with it .
Unless you are buying a used Brocade switch or something, the price does not really compare, also 10GB over CAT6 is not power efficient at all so now your looking at SFP+ (which most people have no idea bout or where to start)
@@TheRealMrGuvernment "most people" arent even looking at 2.5G either. mikrotik has a 4-port sfp+ unit for less than $150, makes it a lot easier to step up to 10G, pcie sfp+ cards and twinax cables are cheap, so's fiber, because theres a premium for everything "Base-T" over 1G
Who's offering a new five port fanless 10 GbE switch for $100?
@@TheRealMrGuvernment You dont need SFP+ for 10G at all . Thats just bs . And even then its not hard at all to do .
@@xythiera7255 I never said you could only use SFP+.....so go back and read again...
Simple economy of scale... there's no market for it yet. Just like 10G, it was very expensive to start with. (if you were around eons ago, so was 100M. and the move from hubs to switches.) Give it time, the market will grow, costs will go down.
my nas is limited by the read/write of a single spinning rust drive so there is no point in upgrading my lan. If I upgraded my nas, I might consider a 2.5gbe but I honestly don’t have a need because I feel like the performance is very good right now.
I built a 8-HDD NAS and it was far from automatic (more RAM, flash cache) to get it to go beyond 2.5gb range, although it was pretty easy to go past 1gb.
A single disk of spinning rust can can transfer at around 180mb/s these days (and faster to/from cache) ... about 65mb/s faster than 1Gbe. I can say I was happy using 2X or even 3X 1Gbe connections to my NAS once SMB3 started giving me the combined bandwidth ... but even happier when I switched to 10Gbe and was able to free up some network ports.
as always, a great review. Even if the switch does not support any vlan mgt, you can ALWAYSs program your NIC in OS to support VLANs, with a VLAN ID. Just moving the management around a little :-)
Totally, but we cannot say that is a feature of the unmanaged switch.
@@ServeTheHomeVideo good call
@@Knightrider159 but the statement remains true either way.... If VLAN manip was available; that would put the switch in the managed platform It is VLAN aware and will move packets based on VLAN ID intra switch (VLAN1)
$10 USD per port is the maximum price for widespread adoption. It's just low enough to be rationalised by the average consumer. For enthusiasts on a budget, this is affordable and they'll not just buy it, they'll recommend it to all their friends. For enthusiasts with more money, 10GbE makes more sense.
I have been looking for a switch with the LEDs separate from the ports. I prefer it since then I can tape over them and don’t need to see the blinking. (For home use).
I went with a MicroTik SFP+ switch instead, crazy how expensive these are for the speed.
I think $100-$120 is a reasonable price that would push mass market adoption. That's inexpensive enough that I could justify throwing that in at my folks house so the handful of devices with higher speed NICs they have could actually have fast access to the NAS I built them
Who are these switches intended for? My only guess is SOHO setups with a 2.5 gigabit NAS. Most consumers are fine with 1gig, and any larger businesses would most likely want 10gig uplink and/or management features. Compared to the Microtik CSS326 I don't see the value in this, but then again I don't have any 2.5gig devices.
I got a new motherboard with 2.5 and now have the itch to use it. There are cable modems with 2.5g now and some consumer routers with 10G wan. Too bad my wish list is nearing 900 bucks which is hard to justify for my casual use and occasional home lab fun
The cost difference to go to 10gb isn't very big. It seems pointless to use 2.5gb unless you already have a bunch of 2.5gb devices