Man, I have the 1g poe version of this switch... it's been rock solid for the last two to three years... I can not justify an upgrade, but it's always good to see Mikrotik pushing forward.
Lovely piece of kit, looking forward to other 2.5gbit devices they're going to produce, I do hope for a middle ground switch between this and the CRS310-8G+2S+IN, like a 16 port or something with 2 or 4 10gbit uplinks. around the 400 dollar mark or so.
Appreciate the update, I was /4 of the way through the original when it was changed out. It is a lot clearer now. However, what I'd really want is to have more 10gig sfp+ rather than the 40gig.
I like it when a company that has high-end products also makes a low-end product. When I have to use something at work, and I can get a cheaper version for my home, it allows more brand loyalty.
If you are only after 40G, go with the quad 100G switch from Mikrotik. Great future proofing, and you can break out the QSFP28 into either 4x10G or 4x25G or 1x40G or 1x100G
That's a bit too pricy for "fututreproofing" my setup. Still decent price nonentheless. You're probably spot-on on fan speculation as there appears to be 4 pin connector for it there. Or maybe they were just planning extended temperature range version or something
The comment on the port selection was exactly what I thought when I looked at it. Cannot figure out why they chose 40Gbps/10Gbps instead of just going for a bunch of 25Gbps. Only logical reason I can think is that they have some kind of limitation with the switch chip. A POE version is essential in the future.
I think you are right that it is a switch chip limitation. Just know at the Marvell Analyst Day 2023, I campaigned for them to get MikroTik new low-cost switch chips.
They could still probably mux it out differently, but cost, and target audience. This screams campus closet switch, where 10Gb SFP+ ports are used for uplinks to campus distribution and the 40G ports used for stacking. Might also be able to swap those if you have 40G Distro, nit sure if those 10G ports are designed for stacking though. I agree that the more ideal next switch design would be predominantly 12-24 1/10Gbase-T access ports, 4-8 mGIG 1G/2.5G/5G POE++ ports, 4-8 10/25G SFP+/SFP28 ports and four 40/100G QSFP/QSFP28 ports. Maximum flexibility, where it could still be deployed as closet stack, but with modern PoE and mGIG for Wi-Fi, optional 1/10Gbase-T to client ports, ability to connect 10/25G devices like modern edge appliances, servers and decent ports for stacking and uplinks. I know I'm asking for too much flexibility.
@@marcogenovesi8570 yeah, I don't know much about the chip they use in this, but some of modern mid-cost Chinese switches seem to be using the chips that are 100G than can mux out to 10x10G / 2x40G+2x10G or 4x25G or even a bunch of mixed 2.5G/5G mGIG. For this price point, it's more like you're getting the beginnings of a more professional switch at prosumer prices because they're using 5yr+ old surplus components, but it's higher quality than some of the newer crap on aliexpress.
Bigger... sure... but not that big. 12 to 16 x 2.5GBe ports with PoE+, quiet or fanless and with at least a handfull of L3 capability, under AUD$400 at the door would be nice. Seems often forgotten that in the real world people are going to have a mix of techm we're not all building high end home labs from scratch with giant budgets. My sticking points... 1) is a NAS I have that has 2 x 1GBe ports where with my current switch has 24 ports (only 1/2 populated) and I'm running link aggregation to get a 2GBe link to the nas. 2) is 8 ports leaves me to short, but 24+ ports is way overkill. 3) Fan noise has given me tinnitus over the years (used to have a desk next to a server rack before anyone cared).
i just got a 2.5 plus 1 10spf - suits my needs and cheap too - could not hold off any longer and will future proof me for a while - the 2 40g on this one are nice for ws to nas connection - hopefully they will be able to add more 40g ports in next few gen of switches - smb mkt needs fast networking - it is their weakest link arguably. with lots of 2.5 you can do bonded connections a bit easier
Really cool switch, I just recently bought the 8-port little brother of this thing and love it. I wish they'd put a bit more grunt into the processor when they've got this many ports and they're all at least 2.5G; you'd have trouble pushing the internal 1G line rate (let alone 2.5G like the external ports) through even a single port if you're doing any sort of routing or filtering, so you're basically forced to leave everything in hardware offload mode and don't really get to do much with the breadth of RouterOS features other than possibly running some services on the built-in processor.
I noticed that the board has four headers for fans. Two are plugged in (FAN 3 & 4), one on the right has a plug for a fan available (FAN 1), and in the middle was another fan port titled FAN 2, but missing the plug. Any chance you could use a multimeter to see if the fan 2 port is alive?
Great overview. I got CRS504-4XQ-IN (4x100G) few months ago, and while it is nice, small and light, low power, and quite, it has troubles. I am not sure I would recommend it (crashes, booting issues, not great performance, issues with channelizing ports, diagnostic, confusing settings, that cannot be change, easy to create loops, i.e. when one also uses management port, poor documentation, interface that saw no improvements in last 20 years, etc). The previous time I used mikrotik was like 15 years ago, mostly small access points, and they were doing fine. I also had their 4x10G mini switch, which was a the best deal at the time (still is great, but I do not use it at the moment, as I updated from 10G to 25/40/100G mostly). 40G is basically dead. I would not buy an yswitch specifically because it has 40G, but at least you can split it into 4x10G, or get cheap NICs for it. As of this 2.5G switch. I guess it is ok. I think fascination with 2.5G is weird, but I guess anything better than 1G is better to move industry forward. And we were stuck on 1G for way too long. If we can get 2.5G for the same price we do 1G, then sure, I am all for it, for it to be a new base. But this QNAP switch is probably more interesting, and better long term. CRS326 4C+20G+2Q+RM, maybe if it would like 400$, then it would be ok. Also love the cooling solution on this guy.
I think 2.5G is the future for most homeowners that have bigger needs. IE for more wifi uplink on the new Wifi 7 stuff. Most don't need even 1G, but 2.5gbps is so cheap that your dumb switches should just support it IMO.
I want a switch with 16x 2.5GbE, 4-8x 10GbE (NOT using expensive and hot SFP+ to copper adapters), 4x SFP+, and a couple 25Gbps uplinks, and which doesn't use crazy amounts of power. I don't like the Mikrotik always wants us to pay sooo much extra for the adapters, which stick way out from the switch, don't cool well (because they can't put nice heatsinks on SFP+ to copper adapters) and just are a pain.
Would be nice for a collapsed core setup for home labbers or power users since it should support MLAG. The price though... If they added 802.3AT on the first 8 ports or actually have 10G multigig on the combo Eth ports, perhaps it'd be alright for that price.
For switches like that, with so much empty space, I wonder why they use those noisy fans, instead larger blower style fans that can run at a lower speed. Think some of the videocards from the 2008-2012 era where they started to use larger blowers to keep noise levels down.
Simple. Where would a blower fan pull air from? Blower fans pull air from above or below. If this thing is in a rack, then even if there are vent holes.. Well, you'd have to leave at least 1U of space open so it could pull in air. Might be fine for you, but that could mean a DOA product for SMBs who could be interested in this, and value rack space.
@@jorper2526I was thinking a half height blower, with a duct pulling in air from the side or back, then sending some airflow to the PSU, and other airflow over the other heatsinks (split flows were done with some cards that used separate heatisnks for the VRAM and VRMs). While some efficiency is lost from a change in air direction, large diameter blowers are pretty quiet while having a decent airflow and static pressure. The PS5 also uses a similar airflow design when the side panels are on.
Will you be releasing a review anytime soon of the Qnap 10G switch shown towards the end of this video (QSW-M3216R-8S8T)? Many thanks for all your informative videos, really helpful when deciding on new tech. 🏆
You might be able to see it on the screen when I talk about the management, but the QSFP+ ports are labeled for each of the four 10Gbps link groups already
passive radiator is for the switch chip, fans are for the PSU. Originally it had a fan for the switch chip too as shown with the missing fan but they decided to save the cost of that fan
I still dont get the logic of combo ports. They are wasted space. Just use RJ45 SFP+. This switch has no PoE, so this would be the only function a SFP+ cant do.
Thanks for the review... I did like and also dislike this switch for few reasons: (1) It has 128 MB of RAM compared to the old CRS326-24G which is a downgrade that has 512 MB ( I found that 64 MB causes random reboots when memory saturates). (2) Flash Storage is 32 MB (double than CRS326-24G) but we recent package upgrades are eating this small storage (why not go 1 GB for RAM And 1 GB for Storage? don't get it and don't tell me is price (3) No PoE++ as you mentioned, so its a downgrade in certain areas compared with older models and price is 3X times higher than the current CRS326-24G which is a DoA.
I paid almost next to nothing on my CRS326-24G-2S+ before the pandemic happened. Also, same for my CRS328-24P-4S+. Chip shortages had huge influence on prices. It's coming back down but nowhere near the prices 4 years ago.
@@Darkk6969True but getting 2.5 Gbps + 2 extra SFP+ + 2 x 40 Gbps for 3x de extra cost of the CRS326-24G, doesn't make much sense to me, specially if they lack 1/5 of the memory of the previous generation
limitations in the max size supported by the very old management CPU (from 2013), which is due to price and availability issues. Also the reason why it uses DDR2 RAM which is old. Same reason it has 40Gbit and not 25/50Gbit ports, they use a switch chip that is old and does not support more modern modes.
Great device! I am using mikrotiks myself since 2010. Only, Mikrotik company coming from an not english speaking country, the name should not be pronounced as Mikrotik (as in the video and for english native speaker) but Meekrotik. John is john everywhere you go. :)
Problem with the MT is that feature set is misleading. Majority of times the feature set is not hardware offloaded and you get a set of ports driven by software on under powered CPU and that software tends to be broken itself.
Patrick, first of all thank you for this review, I am looking for a new managed switch and I am looking on the Mikrotik website for some new switches/routers. This switch is interesting, but I am looking for a 10 GBe variant with more then the standard 8 ports (16 is fine) and maybe some extra fun things like one or two 25GBe or 40GBe ports like in this review. I don't care about PoE, but definitely power consumption/noise. You don't have to break any NDA's, but can you help me by telling me if I should wait for some unreleased products/reviews or should I bite the bullet now and get something that is available at the moment?
@@jorper2526 VLANs, LACP, maybe for some ports more advanced switching features but nothing crazy. The newest Mikrotiks also have hardware L3 routing and switching which is a nice bonus. I don't really need a crazy router or switch, but just something that has some more future proof features, which in my opinion 10 gbe is.
@@masterTigress96 Just keep in mind that any routing is going to hit that (Very slow) management chip, dropping speed below 1gbe. So that means if you are going from VLAN to VLAN it will be slow. Lots of home users fall victim to this.. Just remember, there's a reason why the big enterprise switches that are very loud, and very power hungry, and very expensive.. They sometimes have entire Xeon chips in them just to handle the routing on L3.
@@jorper2526 Agree, but my main concern is noise as I am looking for a router/switch for my homelab. I don't have a garage or something to stuff it away or else I would have opted for used Juniper/Cisco/Aruba switches. I am aware of the limitations of e.g. the Mikrotik hardware but it is the way it is.
Does this switch support Link Aggregation?? If it does i think i found my switch for my homelab/home networking. plug the NAS into the 2 40Gbps ports in a LAG for max performance, then connect the Stream PC and the Gaming PC to the 10G ports in another LAG then the rest of the house can run on the 2.5G network.
Mikrotik has not been very clear about this switch. Their website (and the original video) mentioned that the combo Base-T ports were 10GE but they're actually 2.5G - this looks to have been corrected EDIT: You can find the changes to the video at 2:45
It would be nice to have all 10G posts. I know its a lot more power, but thats ok. I have a MokerLink 2.5G. I wanted the 10G but its $180CAD, then Id have to buy 6 SFP RJ45, and that would already be out of budget.
Lets examine this idea shall we. All of the ports combined require 170gbps switching capacity to be non-blocking. Switching those 2.5gb ports for 10gb would take a chip that can handle 200gbps. After adding the rest of the 10gb and 40gb.. You would need 320gbps switching capacity to be non-blocking. That's almost double. So now, you are adding in a more powerful switch chip, that will draw more power and product more heat. Now you need to redesign the cooling solution to accommodate that, so that means even more cost. Oh, and you'll need to upgrade the power supply for the units.. So more cost. You are left with a switch that is well over $1000, and you're complaining about having to buy 6 SFP modules? :P
I don't get the pursuit of 2.5Gb. It isn't common from my experience. As far as I can tell the cards aren't any cheaper than 10Gb especially when buying used. I wish the standard would die off personally. It's just creating more stuff people wont have a use for in a short period pf time. 10Gb is super cheap as all the datacenters are abandoning it for faster stuff. I can't even sell a used 10Gb switch to save my life unless I just give it away.
The answer is really simple. 2.5GbE can work on the same legacy cabling that is installed in walls. So it’s the zero friction upgrade. It will be a lot more common this year
"As far as I can tell the cards aren't any cheaper than 10Gb especially when buying used." Well, yeah. But go look at new. Used 10G cards also create more heat. And it's not like you can throw a X540 into your AP either. Plenty of stuff is using 2.5g these days as well. "I can't even sell a used 10Gb switch to save my life unless I just give it away." Alright, what do you have and how much are you asking?
@@ServeTheHomeVideo it will be nice if it becomes more mainstream. I have been wanting to see how much of the 10Gb speed I can push through the cat 5 runs in my office. 😁 I feel like the only people needing above gigabit are homelabbers and maybe some more advanced offices.
If you wire your home I think most of your devices has 2.5Gb or less on the motherboard, if you only use a switch in a rack it's a different story. I have 15 devices in my home outside the server rack and only 4 of them are faster than 1G/bits so in that case 2.5 is an good compromise, then the question is if you need 10G/bits for your clients?
@@fordonmekochgalenskaper5665 honestly I messed up and didn't run conduit in my house for all my low voltage for future proofing reason. Most people won't constantly need access to 10Gb let alone 2.5Gb. The jump to 2.5Gb is borderline pointless though. Especially when you factor the price difference between 2.5Gb and 1Gb. Also haven't tried it but you should be able to push 10Gb on lower end cabling depending on the run length and environment I'm told.
...one addl comment. Those QNAP switches ... are just MEH. Seriously meh. The Zyzel 1930 line is SO much better, and you get 10G ports(10/100/1G/2.5/5G/10G) across the board for the same price per port. And a much better and functional webUI. Not even close. I have both, I know :-)
Keep in mind that with combo ports, you don't even have to buy an SFP+ module. Those add up quick. If with the TP-Link you go with Copper SFP+ then that easily adds $200. Those two 40Gb ports have 80% of the bandwidth that entire TP-Link has. If, after buying SFP+ modules you are only spending another $200 for that much more throughput, and you need it.. Well, that might not be a bad deal.
Almost no one cares about 2.5 GbE, I don't know why people keep trying to push it. Anyone who cares about network performance wants 10gig as a minimum. Anyone who doesn't care isn't going to spend the extra cash for a minor upgrade. if the industry isn't going to provide an affordable and reliable 10gig copper solution people will either utilize fiber or continue to not care about these half hearted attempts at progress. The window of opportunity for 2.5GbE was 10 years ago. It's time to move on.
You're missing the point of 2.5GbE. 10GbE over twisted pair was first defined as a standard in 2006, almost 20 years ago. Even today, it remains a rare feature on consumer grade hardware. Most consumer electronics stuck with 1000 Base-T, a standard dating back to the previous century. The fact remains that 10G over twisted pairs is too costly and too hot to replace 1000 Base-T, and moreover it requires higher-standard cabling. 2.5GbE is a standard first defined in 2016, precisely because of the troubled adoption of 10GbE over twisted pairs. Importantly, it is certified for the same cabling standards as the 1000 Base-T it is meant to replace. And speaking from personal experience, I've never had thermal issues with 2.5GbE the way I had with 10G Base-T (the 10G SFP+ to RJ45 transceiver I had plugged into a cable closet switch would frequently throttle down to < 100Mbps with severe packet losses and latency fluctuations. It was bad enough for me to go back to 1G) TLDR: 2.5GbE is positioned to replace the ubiquitous 1000 Base-T in nearly all use cases. It's not the fanciest of standards, but it is what the mainstream will look like in a few years.
@@DGao-zz5vqThank you for taking the time to reply. There are certainly many issues with the current implementations of 10GbE and I can see how 2.5GbE looks attractive compared to it. As far as I'm aware there still is not a single implementation of a reliable 10GbE Thunderbolt adapter. My comments are less towards how these 2.5GbE products compare to what you can buy in the 10GbE space, and more towards the sentiment that we shouldn't be happy with table scraps. Many of us in the comment section making similar remarks aren't missing the point. We're well aware that there's plenty of CAT5/6 installations in place and operational. We're criticizing the bigger picture issues about how the state of BaseT networking even got this bad in the first place. We have wireless solutions in the market today that mop the floor of current mainstream wired copper connections, 2.5GbE included. These Wifi 7 implementations are in phones, which are far more difficult to manage thermal and power constraints. Heat and power problems on network adapters for laptop and workstations is an issue that modern process technologies have brute forced long ago. If the current chip sets are problematic than it's a design and node selection issue. The 2.5 band-aid is also a solution in search of a problem. Any organization that's feeling pain points due to networking speeds isn't going to rip out their entire network stacks for a 2.5x performance jump. They'll replace their wiring and support 10gig or go with fiber. A competent 10 gig solution would fallback to 2.5/5GbE if it can't negotiate at 10gig due to weak wiring. I expect the consumer space to embrace Wi-Fi as a 'good enough' solution, it's actually damn impressive now. I never thought I'd see the pro-sumer space implicitly sabotage the progress of networking connectivity by pushing half-baked wired solutions that are inferior to wireless solutions in every way. The 2.5/5GbE standards are great as fallbacks for when endpoints and wiring aren't up to 10gig levels, but I don't see any place for NICs that are 2.5GbE other than 'that's what happened to come with the product I bought from Costco' A 2.5X performance increase over 25 years is nothing to get happy about. Could you imagine if the PCIE adoption stagnated this badly, we'd still be on PCIE 2.0.
I still don't get why QSFP 40G, that's such a dead end at this price point and for modern uplink connectivity. 10, 25, or 100 are the speeds. 40 is such an odd bastard that's incompatible with 25/100, why why why Mikrotik?
The pricing is what kills it for me. At $999 this is a hard nope. Even at $800 it's still a no from me dawg. If it had 2x25G SFP+ and a $499 price tag I'd have ordered two of them already. It just doesn't offer enough over a pair of their own CRS310 8-port switches to justify the cost. Frankly, I don't understand why they even launched this product. 40G is obsolete even for homelabbers, now that 25G has gotten super affordable.
real switches belong in caged racks in data centers or urban office buildings with conditioned power controlled access, drug sniffing dogs and armed guards. .real people at home (or free on bail) before & after work don't need to permit packets to propagate on their premises.
It's peace of crap, not based upon a quality but idea. We (as consumers) are waiting 24/48 port 10G switch for the last 3+ years. What should i do with 2.5G, plug in AP's. As they did not have idea or some kind of cpu/architecture limitation!?
10Gbase-T is very challenging for the market because of limitations running over existing cabling installed in walls. 2.5GbE is a good upgrade over 1GbE and does not have this issue. Another challenge is that for higher-speed switch silicon, everything is going to 25GbE/ 100GbE instead of 10GbE/ 40GbE because 10GbE is very inefficient for PCIe lane use. Later this year and into next year, a lot of the DC gear is going to 50GbE and 100GbE being the lower-end standard speeds. So perhaps the challenge is that at the bottom end, people who what a cheap upgrade on existing wiring (2.5GbE), and at the higher-end (servers) people left 10GbE networking 5-7 years ago.
Man, I have the 1g poe version of this switch... it's been rock solid for the last two to three years... I can not justify an upgrade, but it's always good to see Mikrotik pushing forward.
Try the same spoofing. Make sure that it is all on the processor.
Lovely piece of kit, looking forward to other 2.5gbit devices they're going to produce, I do hope for a middle ground switch between this and the CRS310-8G+2S+IN, like a 16 port or something with 2 or 4 10gbit uplinks. around the 400 dollar mark or so.
Appreciate the update, I was /4 of the way through the original when it was changed out. It is a lot clearer now. However, what I'd really want is to have more 10gig sfp+ rather than the 40gig.
QSFP+ to SFP+ converters will get you two more.
You mean 8 more :) Each QSFP+ should give you 4 SFP+ with breakout cables@@RambozoClown
Love my Mikrotik 1/10/40 switch this is definitely the switch I would have prefered and might need to check it out after it comes back down a bit.
I like it when a company that has high-end products also makes a low-end product. When I have to use something at work, and I can get a cheaper version for my home, it allows more brand loyalty.
There is even a populated 4pin FAN header right next to "missing" FAN :)
And the thing to add... WinBox also runs with Wine on Linux.
Great point
And Mac OS X
@@kwinsch7423 fair, but I wouldn't use apple, since they don't take it seriously with their "Your data is your data" stuff. But otherwise an ok os.
Now has a native version
Literally was looking at 40G switches yesterday.
If you are only after 40G, go with the quad 100G switch from Mikrotik. Great future proofing, and you can break out the QSFP28 into either 4x10G or 4x25G or 1x40G or 1x100G
I was watching this when first one was pulled; what’s going on to make a rerelease?
I'm sorry but does that switch have a mullet?!
Ha. Good one!
Well "business in the front, party in the back" does describe it pretty well
@@ServeTheHomeVideovbj
What I most like about this switch is the dinosaur scales Mikrotik uses for cooling at the back.
:)
You just made it sound even cooler
11:17 actually, if I'm not mistaken, Mikrotik devices broadcast, the application doesn't scan.
yup, AFAIK they "announce" themselves on all ports you configured for the neighbors feature (which by default are usually all).
That's a bit too pricy for "fututreproofing" my setup. Still decent price nonentheless.
You're probably spot-on on fan speculation as there appears to be 4 pin connector for it there. Or maybe they were just planning extended temperature range version or something
The comment on the port selection was exactly what I thought when I looked at it. Cannot figure out why they chose 40Gbps/10Gbps instead of just going for a bunch of 25Gbps. Only logical reason I can think is that they have some kind of limitation with the switch chip.
A POE version is essential in the future.
I think you are right that it is a switch chip limitation. Just know at the Marvell Analyst Day 2023, I campaigned for them to get MikroTik new low-cost switch chips.
the switch chip is an old design and can't do that
They could still probably mux it out differently, but cost, and target audience. This screams campus closet switch, where 10Gb SFP+ ports are used for uplinks to campus distribution and the 40G ports used for stacking. Might also be able to swap those if you have 40G Distro, nit sure if those 10G ports are designed for stacking though.
I agree that the more ideal next switch design would be predominantly 12-24 1/10Gbase-T access ports, 4-8 mGIG 1G/2.5G/5G POE++ ports, 4-8 10/25G SFP+/SFP28 ports and four 40/100G QSFP/QSFP28 ports. Maximum flexibility, where it could still be deployed as closet stack, but with modern PoE and mGIG for Wi-Fi, optional 1/10Gbase-T to client ports, ability to connect 10/25G devices like modern edge appliances, servers and decent ports for stacking and uplinks. I know I'm asking for too much flexibility.
@@racerex340 they can't mux it, the switch chip does not support 25gbit because it's old design
@@marcogenovesi8570 yeah, I don't know much about the chip they use in this, but some of modern mid-cost Chinese switches seem to be using the chips that are 100G than can mux out to 10x10G / 2x40G+2x10G or 4x25G or even a bunch of mixed 2.5G/5G mGIG. For this price point, it's more like you're getting the beginnings of a more professional switch at prosumer prices because they're using 5yr+ old surplus components, but it's higher quality than some of the newer crap on aliexpress.
Bigger... sure... but not that big.
12 to 16 x 2.5GBe ports with PoE+, quiet or fanless and with at least a handfull of L3 capability, under AUD$400 at the door would be nice.
Seems often forgotten that in the real world people are going to have a mix of techm we're not all building high end home labs from scratch with giant budgets.
My sticking points... 1) is a NAS I have that has 2 x 1GBe ports where with my current switch has 24 ports (only 1/2 populated) and I'm running link aggregation to get a 2GBe link to the nas.
2) is 8 ports leaves me to short, but 24+ ports is way overkill.
3) Fan noise has given me tinnitus over the years (used to have a desk next to a server rack before anyone cared).
Can't wait for more new MikroTiks.
The team and I are recording two MikroTik videos this week.
i just got a 2.5 plus 1 10spf - suits my needs and cheap too - could not hold off any longer and will future proof me for a while - the 2 40g on this one are nice for ws to nas connection - hopefully they will be able to add more 40g ports in next few gen of switches - smb mkt needs fast networking - it is their weakest link arguably. with lots of 2.5 you can do bonded connections a bit easier
Awesome, thanks for the demo and info
Mikrotik is Awesome!!
Very
This thing looks insane
Very insane!
Really cool switch, I just recently bought the 8-port little brother of this thing and love it. I wish they'd put a bit more grunt into the processor when they've got this many ports and they're all at least 2.5G; you'd have trouble pushing the internal 1G line rate (let alone 2.5G like the external ports) through even a single port if you're doing any sort of routing or filtering, so you're basically forced to leave everything in hardware offload mode and don't really get to do much with the breadth of RouterOS features other than possibly running some services on the built-in processor.
Spot on
I noticed that the board has four headers for fans. Two are plugged in (FAN 3 & 4), one on the right has a plug for a fan available (FAN 1), and in the middle was another fan port titled FAN 2, but missing the plug. Any chance you could use a multimeter to see if the fan 2 port is alive?
I would love to see a 2.5G active PoE switch! You can't seem to get them anywhere
Any chance of an upcomming switch with all 10gbit ports + sfp28 or higher?
Great overview.
I got CRS504-4XQ-IN (4x100G) few months ago, and while it is nice, small and light, low power, and quite, it has troubles. I am not sure I would recommend it (crashes, booting issues, not great performance, issues with channelizing ports, diagnostic, confusing settings, that cannot be change, easy to create loops, i.e. when one also uses management port, poor documentation, interface that saw no improvements in last 20 years, etc).
The previous time I used mikrotik was like 15 years ago, mostly small access points, and they were doing fine. I also had their 4x10G mini switch, which was a the best deal at the time (still is great, but I do not use it at the moment, as I updated from 10G to 25/40/100G mostly). 40G is basically dead. I would not buy an yswitch specifically because it has 40G, but at least you can split it into 4x10G, or get cheap NICs for it.
As of this 2.5G switch. I guess it is ok. I think fascination with 2.5G is weird, but I guess anything better than 1G is better to move industry forward. And we were stuck on 1G for way too long. If we can get 2.5G for the same price we do 1G, then sure, I am all for it, for it to be a new base.
But this QNAP switch is probably more interesting, and better long term.
CRS326 4C+20G+2Q+RM, maybe if it would like 400$, then it would be ok.
Also love the cooling solution on this guy.
I think 2.5G is the future for most homeowners that have bigger needs. IE for more wifi uplink on the new Wifi 7 stuff. Most don't need even 1G, but 2.5gbps is so cheap that your dumb switches should just support it IMO.
The rear heat sink looks like the Cisco Catalyst 40Gbe QSFP module covers
I want a switch with 16x 2.5GbE, 4-8x 10GbE (NOT using expensive and hot SFP+ to copper adapters), 4x SFP+, and a couple 25Gbps uplinks, and which doesn't use crazy amounts of power.
I don't like the Mikrotik always wants us to pay sooo much extra for the adapters, which stick way out from the switch, don't cool well (because they can't put nice heatsinks on SFP+ to copper adapters) and just are a pain.
inb4 unifi releases one with rgb for 2000
This will make sure bi pi-hole blocks ads at a blazingly fast speed.
Would be nice for a collapsed core setup for home labbers or power users since it should support MLAG.
The price though... If they added 802.3AT on the first 8 ports or actually have 10G multigig on the combo Eth ports, perhaps it'd be alright for that price.
For the price I'll just have to stick with my cisco c9300-24ux switch. I know it pulls more power, but it's mgig full and poe.
Name another mikrotik switch with 40g and Poe+ ???
For switches like that, with so much empty space, I wonder why they use those noisy fans, instead larger blower style fans that can run at a lower speed. Think some of the videocards from the 2008-2012 era where they started to use larger blowers to keep noise levels down.
Simple. Where would a blower fan pull air from? Blower fans pull air from above or below. If this thing is in a rack, then even if there are vent holes.. Well, you'd have to leave at least 1U of space open so it could pull in air. Might be fine for you, but that could mean a DOA product for SMBs who could be interested in this, and value rack space.
@@jorper2526I was thinking a half height blower, with a duct pulling in air from the side or back, then sending some airflow to the PSU, and other airflow over the other heatsinks (split flows were done with some cards that used separate heatisnks for the VRAM and VRMs). While some efficiency is lost from a change in air direction, large diameter blowers are pretty quiet while having a decent airflow and static pressure.
The PS5 also uses a similar airflow design when the side panels are on.
I desperately want a micro til 12 Port poe++ with 10 gig ports.
Will you be releasing a review anytime soon of the Qnap 10G switch shown towards the end of this video (QSW-M3216R-8S8T)?
Many thanks for all your informative videos, really helpful when deciding on new tech. 🏆
Can QSFPs be broken into 4x10G ports? Or the are fixed to a single port?
I know there are (at least) qsfp4x sfp DACs. Not sure about other physical connections
You might be able to see it on the screen when I talk about the management, but the QSFP+ ports are labeled for each of the four 10Gbps link groups already
how do you even utilize 40g from storage
Poe version would be very useful!
These names are getting out of hand. At this rate we are going to need higher resolution screens just to fit the name into the video
Why both passive radiator and fans in the same device? Can’t be an economically wise decision
passive radiator is for the switch chip, fans are for the PSU. Originally it had a fan for the switch chip too as shown with the missing fan but they decided to save the cost of that fan
@@marcogenovesi8570 to my knowledge a big custom made radiator with custom made heat pipes is far more expensive than a regular heatsink and a fan.
You got a 1u 10gb poe switch managed for a good price?
I personally dont't like combo ports because you have to pay for what you can't use. 10gb ethernet is much more expensive than sfp+, isn't it?
I still dont get the logic of combo ports. They are wasted space. Just use RJ45 SFP+. This switch has no PoE, so this would be the only function a SFP+ cant do.
Thanks for the review... I did like and also dislike this switch for few reasons: (1) It has 128 MB of RAM compared to the old CRS326-24G which is a downgrade that has 512 MB ( I found that 64 MB causes random reboots when memory saturates). (2) Flash Storage is 32 MB (double than CRS326-24G) but we recent package upgrades are eating this small storage (why not go 1 GB for RAM And 1 GB for Storage? don't get it and don't tell me is price (3) No PoE++ as you mentioned, so its a downgrade in certain areas compared with older models and price is 3X times higher than the current CRS326-24G which is a DoA.
I paid almost next to nothing on my CRS326-24G-2S+ before the pandemic happened. Also, same for my CRS328-24P-4S+. Chip shortages had huge influence on prices. It's coming back down but nowhere near the prices 4 years ago.
@@Darkk6969True but getting 2.5 Gbps + 2 extra SFP+ + 2 x 40 Gbps for 3x de extra cost of the CRS326-24G, doesn't make much sense to me, specially if they lack 1/5 of the memory of the previous generation
limitations in the max size supported by the very old management CPU (from 2013), which is due to price and availability issues. Also the reason why it uses DDR2 RAM which is old.
Same reason it has 40Gbit and not 25/50Gbit ports, they use a switch chip that is old and does not support more modern modes.
Dangit, I just upgraded to 10 gig last month.
Great device! I am using mikrotiks myself since 2010. Only, Mikrotik company coming from an not english speaking country, the name should not be pronounced as Mikrotik (as in the video and for english native speaker) but Meekrotik. John is john everywhere you go. :)
Problem with the MT is that feature set is misleading. Majority of times the feature set is not hardware offloaded and you get a set of ports driven by software on under powered CPU and that software tends to be broken itself.
Patrick, first of all thank you for this review, I am looking for a new managed switch and I am looking on the Mikrotik website for some new switches/routers. This switch is interesting, but I am looking for a 10 GBe variant with more then the standard 8 ports (16 is fine) and maybe some extra fun things like one or two 25GBe or 40GBe ports like in this review.
I don't care about PoE, but definitely power consumption/noise. You don't have to break any NDA's, but can you help me by telling me if I should wait for some unreleased products/reviews or should I bite the bullet now and get something that is available at the moment?
I would say first off, what do you want management for? What features do you want?
@@jorper2526 VLANs, LACP, maybe for some ports more advanced switching features but nothing crazy. The newest Mikrotiks also have hardware L3 routing and switching which is a nice bonus. I don't really need a crazy router or switch, but just something that has some more future proof features, which in my opinion 10 gbe is.
@@masterTigress96 Just keep in mind that any routing is going to hit that (Very slow) management chip, dropping speed below 1gbe. So that means if you are going from VLAN to VLAN it will be slow. Lots of home users fall victim to this..
Just remember, there's a reason why the big enterprise switches that are very loud, and very power hungry, and very expensive.. They sometimes have entire Xeon chips in them just to handle the routing on L3.
@@jorper2526 Agree, but my main concern is noise as I am looking for a router/switch for my homelab. I don't have a garage or something to stuff it away or else I would have opted for used Juniper/Cisco/Aruba switches. I am aware of the limitations of e.g. the Mikrotik hardware but it is the way it is.
I don't even /need/ this thing - but I WANT ONE...
Nice, sadly not in my budget yet.
With POE it would be the perfect switch.
truely a shame it doesn't have PoE.
I totally agree
Does this switch support Link Aggregation?? If it does i think i found my switch for my homelab/home networking. plug the NAS into the 2 40Gbps ports in a LAG for max performance, then connect the Stream PC and the Gaming PC to the 10G ports in another LAG then the rest of the house can run on the 2.5G network.
I might be wrong, but all my Mikrotik’d router/switches can do LAG (using it actualy)
@@ReneGo-that's good to know, might look at investing in this switch, if it can do LAG then might as well jump into it
Did you tried put a load on it?
We showed this.
@@ServeTheHomeVideo it wasn't really load. I mean populate all ports and generate traffic. RouterOS isn't good for such use case.
Could you guys consider testing the Unifi Pro Max 24 switch? I’ve been considering it as I’m already in the Unifi ecosystem
I just got a pro max switch. It’s definitely quirky and I have a few things I love and hate about it.
Power this switch off poe++, I have a little bitty mikrotic with like 3 or 4 ways to power it, redolently.
TIL there's still chip makers producing ddr2
there is a lot of old stock. The management chip is from 2013 and is also old stock
I just want an unmanaged 16 port 2.5G switch... Some 10GBASE-T would be nice too, but even just a 16 port 2.5 would be great.
No longer early this time. :( What was the correction?
i wonder aswell, maybe theys got smth wrong and had to reupload it to correct details
Mikrotik has not been very clear about this switch. Their website (and the original video) mentioned that the combo Base-T ports were 10GE but they're actually 2.5G - this looks to have been corrected
EDIT: You can find the changes to the video at 2:45
Mikrotik is just like early days of Linux as OS: a bunch of parts that combined together, works just fine. Looks like garbage, but works just fine.
ua-cam.com/video/tMrzdKzQTf8/v-deo.html
It would be nice to have all 10G posts.
I know its a lot more power, but thats ok.
I have a MokerLink 2.5G. I wanted the 10G but its $180CAD, then Id have to buy 6 SFP RJ45, and that would already be out of budget.
Lets examine this idea shall we.
All of the ports combined require 170gbps switching capacity to be non-blocking.
Switching those 2.5gb ports for 10gb would take a chip that can handle 200gbps.
After adding the rest of the 10gb and 40gb.. You would need 320gbps switching capacity to be non-blocking. That's almost double. So now, you are adding in a more powerful switch chip, that will draw more power and product more heat. Now you need to redesign the cooling solution to accommodate that, so that means even more cost. Oh, and you'll need to upgrade the power supply for the units.. So more cost.
You are left with a switch that is well over $1000, and you're complaining about having to buy 6 SFP modules? :P
I don't get the pursuit of 2.5Gb. It isn't common from my experience. As far as I can tell the cards aren't any cheaper than 10Gb especially when buying used. I wish the standard would die off personally. It's just creating more stuff people wont have a use for in a short period pf time. 10Gb is super cheap as all the datacenters are abandoning it for faster stuff. I can't even sell a used 10Gb switch to save my life unless I just give it away.
The answer is really simple. 2.5GbE can work on the same legacy cabling that is installed in walls. So it’s the zero friction upgrade. It will be a lot more common this year
"As far as I can tell the cards aren't any cheaper than 10Gb especially when buying used."
Well, yeah. But go look at new. Used 10G cards also create more heat. And it's not like you can throw a X540 into your AP either. Plenty of stuff is using 2.5g these days as well.
"I can't even sell a used 10Gb switch to save my life unless I just give it away."
Alright, what do you have and how much are you asking?
@@ServeTheHomeVideo it will be nice if it becomes more mainstream. I have been wanting to see how much of the 10Gb speed I can push through the cat 5 runs in my office. 😁 I feel like the only people needing above gigabit are homelabbers and maybe some more advanced offices.
If you wire your home I think most of your devices has 2.5Gb or less on the motherboard, if you only use a switch in a rack it's a different story. I have 15 devices in my home outside the server rack and only 4 of them are faster than 1G/bits so in that case 2.5 is an good compromise, then the question is if you need 10G/bits for your clients?
@@fordonmekochgalenskaper5665 honestly I messed up and didn't run conduit in my house for all my low voltage for future proofing reason. Most people won't constantly need access to 10Gb let alone 2.5Gb. The jump to 2.5Gb is borderline pointless though. Especially when you factor the price difference between 2.5Gb and 1Gb. Also haven't tried it but you should be able to push 10Gb on lower end cabling depending on the run length and environment I'm told.
Would be really excited for a PoE version at around the same msrp!
If this was 5 GbE ports instead of 2.5 this would interest me. Also a model with PoE.
5GbE pricing is so close to 10Gbase-T multi-gig that may as well look for 10Gbase-T
@@ServeTheHomeVideo That's not the case at all, plus I have zero use for 10GbE for APs.
...one addl comment. Those QNAP switches ... are just MEH. Seriously meh. The Zyzel 1930 line is SO much better, and you get 10G ports(10/100/1G/2.5/5G/10G) across the board for the same price per port. And a much better and functional webUI. Not even close. I have both, I know :-)
Mikrotik likes to reuse case hardware, so if the want a fan later you don’t need to retool
I think that is what is going on here.
ehh just looks like mikrotik is getting rid of old silicon
I think they probably get Marvell's older switch silicon at much better prices than the 25GbE/ 100GbE stuff.
TP-Link TL-SG3428X-M2 has 24x 2.5GbE + 4x 10GbE SFP+ for half the price. Those 40 GbE ports cannot justify the price difference.
Keep in mind that with combo ports, you don't even have to buy an SFP+ module. Those add up quick. If with the TP-Link you go with Copper SFP+ then that easily adds $200.
Those two 40Gb ports have 80% of the bandwidth that entire TP-Link has. If, after buying SFP+ modules you are only spending another $200 for that much more throughput, and you need it.. Well, that might not be a bad deal.
Almost no one cares about 2.5 GbE, I don't know why people keep trying to push it. Anyone who cares about network performance wants 10gig as a minimum. Anyone who doesn't care isn't going to spend the extra cash for a minor upgrade. if the industry isn't going to provide an affordable and reliable 10gig copper solution people will either utilize fiber or continue to not care about these half hearted attempts at progress. The window of opportunity for 2.5GbE was 10 years ago. It's time to move on.
You're missing the point of 2.5GbE.
10GbE over twisted pair was first defined as a standard in 2006, almost 20 years ago. Even today, it remains a rare feature on consumer grade hardware. Most consumer electronics stuck with 1000 Base-T, a standard dating back to the previous century.
The fact remains that 10G over twisted pairs is too costly and too hot to replace 1000 Base-T, and moreover it requires higher-standard cabling.
2.5GbE is a standard first defined in 2016, precisely because of the troubled adoption of 10GbE over twisted pairs. Importantly, it is certified for the same cabling standards as the 1000 Base-T it is meant to replace. And speaking from personal experience, I've never had thermal issues with 2.5GbE the way I had with 10G Base-T (the 10G SFP+ to RJ45 transceiver I had plugged into a cable closet switch would frequently throttle down to < 100Mbps with severe packet losses and latency fluctuations. It was bad enough for me to go back to 1G)
TLDR: 2.5GbE is positioned to replace the ubiquitous 1000 Base-T in nearly all use cases. It's not the fanciest of standards, but it is what the mainstream will look like in a few years.
Do you have a link for a managed switch with 24 10gig-ports for under 500USD?
Exactly. And the big benefit is being able to replace 1G on wire that has been installed in buildings for decades
@@DGao-zz5vqThank you for taking the time to reply. There are certainly many issues with the current implementations of 10GbE and I can see how 2.5GbE looks attractive compared to it. As far as I'm aware there still is not a single implementation of a reliable 10GbE Thunderbolt adapter. My comments are less towards how these 2.5GbE products compare to what you can buy in the 10GbE space, and more towards the sentiment that we shouldn't be happy with table scraps.
Many of us in the comment section making similar remarks aren't missing the point. We're well aware that there's plenty of CAT5/6 installations in place and operational. We're criticizing the bigger picture issues about how the state of BaseT networking even got this bad in the first place. We have wireless solutions in the market today that mop the floor of current mainstream wired copper connections, 2.5GbE included. These Wifi 7 implementations are in phones, which are far more difficult to manage thermal and power constraints. Heat and power problems on network adapters for laptop and workstations is an issue that modern process technologies have brute forced long ago. If the current chip sets are problematic than it's a design and node selection issue. The 2.5 band-aid is also a solution in search of a problem. Any organization that's feeling pain points due to networking speeds isn't going to rip out their entire network stacks for a 2.5x performance jump. They'll replace their wiring and support 10gig or go with fiber. A competent 10 gig solution would fallback to 2.5/5GbE if it can't negotiate at 10gig due to weak wiring.
I expect the consumer space to embrace Wi-Fi as a 'good enough' solution, it's actually damn impressive now. I never thought I'd see the pro-sumer space implicitly sabotage the progress of networking connectivity by pushing half-baked wired solutions that are inferior to wireless solutions in every way. The 2.5/5GbE standards are great as fallbacks for when endpoints and wiring aren't up to 10gig levels, but I don't see any place for NICs that are 2.5GbE other than 'that's what happened to come with the product I bought from Costco'
A 2.5X performance increase over 25 years is nothing to get happy about. Could you imagine if the PCIE adoption stagnated this badly, we'd still be on PCIE 2.0.
$880-999!?!?!?!? I almost fell out my chair. Realtek automotive chip 8 port switches for $250 might be the move.....
That seems reasonable considering it has 2 40gig ports and is new. Under $1000 is what I was hoping for when I saw the thumbnail.
I still don't get why QSFP 40G, that's such a dead end at this price point and for modern uplink connectivity. 10, 25, or 100 are the speeds. 40 is such an odd bastard that's incompatible with 25/100, why why why Mikrotik?
Totally agree. 25/100G require faster per channel speed, but I think they are becoming more useful.
They use old switch chips, which are cheaper
2.5g is nice but we really need an inexpensive option for a 12-16port 10g switch like sub $500
if this switch had 12 ports of 1Gb PoE++ and 12 ports of 2.5GbE, it would be a super bestseller.
NOOOOOOOOOOOOO!!!! no more 2.5gbe! - this should be a 24 port 10g base-t / 10g sfp+ / 40gbe qsfp switch.
True, but then it would use a lot more power.
24port 10gbase-t would need a dedicated diesel power generator
@@marcogenovesi8570Unifi EnterpriseXG has 24 10g base-T, plus two SFP28 ports, and draws a maximum 100 watts. So not so bad.
@@marcogenovesi8570I own a netgear xs728t that works great and doesn't really take up all that much power.
And more heat
Nice redundant power supply gag
The pricing is what kills it for me. At $999 this is a hard nope. Even at $800 it's still a no from me dawg. If it had 2x25G SFP+ and a $499 price tag I'd have ordered two of them already. It just doesn't offer enough over a pair of their own CRS310 8-port switches to justify the cost. Frankly, I don't understand why they even launched this product. 40G is obsolete even for homelabbers, now that 25G has gotten super affordable.
real switches belong in caged racks in data centers or urban office buildings with conditioned power controlled access, drug sniffing dogs and armed guards. .real people at home (or free on bail) before & after work don't need to permit packets to propagate on their premises.
argh, no POE
Fanless adjacent? 😂
غ توم
It's peace of crap, not based upon a quality but idea. We (as consumers) are waiting 24/48 port 10G switch for the last 3+ years. What should i do with 2.5G, plug in AP's. As they did not have idea or some kind of cpu/architecture limitation!?
10Gbase-T is very challenging for the market because of limitations running over existing cabling installed in walls. 2.5GbE is a good upgrade over 1GbE and does not have this issue. Another challenge is that for higher-speed switch silicon, everything is going to 25GbE/ 100GbE instead of 10GbE/ 40GbE because 10GbE is very inefficient for PCIe lane use. Later this year and into next year, a lot of the DC gear is going to 50GbE and 100GbE being the lower-end standard speeds. So perhaps the challenge is that at the bottom end, people who what a cheap upgrade on existing wiring (2.5GbE), and at the higher-end (servers) people left 10GbE networking 5-7 years ago.
I love you videos, but DUDE, PLEASE use a compressor on your mic. Every sentence after a cut in the video is damaging my hearing. Pretty please.
USW-Pro-Aggregation is better product