Great tutorial! My question: My homeserver for backup has 7 drives in the storage space. Total 7 TB. If my motherboard goes kaput, will the new system configure the pool by itself? Or do I need to backup the backup server?
Short answer: no, you do not have to backup the server. Long answer: Configuration of the pool and virtual disks is stored onto disks themselves. That is why "metadata" shaves a little bit off of capacity. If you remove all the drives, and plug them into a new Windows system that supports Storage Spaces, it will recognize configuration of virtual disks and should automagically appear there without any intervention from you, except plugging the drives in.
I've been using Storage Spaces for 8-9 years on a Windows 8 to 10 computer that is being used a server. I have a two-way mirror and simple space set up. In all of that time I've never had a single issue with it. My only gripe is the inability to reduce thin provisioning size and the inability to make non-thin provisioned volumes. Minor quibbles.
Can someone list a few real-world reasons for "show that I have a drive that's much bigger than I really have"? Other than being EXTREMELY misleading... what's the point? The extra space can never be used. Why not keep all the numbers accurate, and then when you really add the new disk *THEN* it will show the extra available space. Why would I want to falsely believe that I have a 4TB drive that is only 50% full, when it's really just a 2TB drive that's 99.99% full? Dangerous.
12:07 The question you ask yourself after using Windows storage space (Edit: too bad current lastest update can't format properly, so I created the storage pool and attempted to create a drive. Its online in disk management, so I delete the volume and format it from there. I don't know if that's a legit way.)
After watch your video a few times, I will keep my: Least critical files on the stripping w/o parity drive Middle critical files on the stripping with parity drive Most critical files on the mirroring drive all across three or more physical disks. Thks
Thanks for an idea. Unfortunately, I have not yet used tiers in Storage Spaces. To achieve high performance, I used multiple columns (configured high "column count" number) when creating virtual disks. That makes disks run in parallel, much faster than single drive. See here: ua-cam.com/video/rl7t8FToDsU/v-deo.html
@@VedranKesegicSQL Thanks for the link! I found this wiki, but not sure why the sizes are marked like they are. wiki.towaso.de/index.php/Create_Tiered_Storage_Space
Also found this, which is what I'm going to try with 2 x SSDs and 3 x HDDs: forum.level1techs.com/t/how-to-set-up-tiered-storage-on-windows-10/126840/20
Thks ever so much, I just ordered a few USB thumbdrives to replicate your 'Storage Spaces' video. If it works half as well, I'll be able to backup my ~40GB data on multiple mirrored thumdrives, keep one in the car, & once ever few days sync the one in the car to the other mirrored thumdrives. Oh, you didn't try reconnecting the failed disk. If you ever update the video, you might try that.
Huh, USB drives were just for demo purpose to be able to pull them out to simulate failure. For actual backup you would be probably better with some cloud provider (eg. google drive, mega.nz, dropbox, onedrive). For bigger sets of data, a NAS is a decent option, even the smallest one like described here: blog.sqlxdetails.com/crypto-virus-resistent-backup/ Storage Spaces are not intended for backup purpose, just for resiliency to disk failure and speed, sort of more flexible RAID. It is not intended to work with one drive "unplugged" for a long periods of time. Might work though, but that is not what it is made for.
@@VedranKesegicSQL Roger-that & thks for the concern, *I have a Z2 Mini G4 right. Also I'll probably get a OWC Mercury Elite Pro Quad & manage it via MSWindow Storage Spaces. Oh, I saving my thunderbolt3 port for an external GPU. HP Z2 Mini workstation PC ua-cam.com/video/qcjNm9i_pws/v-deo.html OWC Mercury Elite Pro Quad hdd bay enclosure (10Gb/s USB 3.1 Gen 2) eshop.macsales.com/item/OWC/MEQCTJBT00/ *Over 6 months ago I was focused on a Synology NAS but their 4&above HDD bay NASes cost an arm&leg as well as contain weak CPUs. Now I plan to keep my mass-storage attached to my workstation & get an inexpensive 2bay Synology NAS mainly for interfacing with/over the internet from my home. Synology DiskStation NAS Introduction ua-cam.com/video/CaInLiK4zFM/v-deo.html Synology - Play. Collaborate. Achieve. Live BeyondCloud ua-cam.com/video/UsnBqL_pzcs/v-deo.html Synology - Global Event Introduction Video ua-cam.com/video/IXPnP__Ipvc/v-deo.html Synology - Be Your Own Cloud ua-cam.com/video/QOtW9CBsTys/v-deo.html Synology | Introduction to Synology Collaboration Suite ua-cam.com/video/M9OZ-i9N8NA/v-deo.html Eventually I'll emulate everything the Synology NAS does for me in Linux on my workstation but that's take a couple of years (I'm a bored retired physicist). Of course *I'll do versioned/near-real-time backups. But I currently have only ~20GB of personnel data I really care about. It would be nice to mirror/ecrypt it to three thumbdrives, normally keep one thumbdrive in my vehicle, & resync that one in my car once or twice a week. When a thumbdrive dies, I just destroy it & get another one. If Storage Spaces won't do it, I try some other way. *I like the simplicity of Storage Spaces though. Let me know what you think.
This is great illustration video thank you, i would like to add that moving storage disks into another system will keep the storage spaces configuration same ... works like a charm. I have been using storage spaces for almost 4 years and i totally in love with it, i custom build my home HTPC with hostswap drives and using storage spaces as media backup and network file-share storage. its a grate solution if you don't want to lock your data with hardware raid motherboards/cards or invest in NAS.
SS is certainly great for a simple 2-drive mirror...; 1 drive crashes/fails, nothing lost at all. (Unless you are OK with very slow writes, I've avoid parity raids with storage spaces, as the write speeds are abysmal!)
Great video man. Thank you for all that information! I have one question please. At the last of the video you remove the USB like a hdd failure. After a while lot of warnings in storage space management. But what we can do after. Let's say that this usb was a hdd that suddenly died! How can we add a new replacement and tell to the storage Spaces to start recreating the data in the new drive? We can't let them like that and work we the warnings right? I have found that we can first remove and the add a drive but I don't know how we can do that when some of our drives is died. No option for remove imas it seems in your video. Thank you man again
If a disk dies, you can add a new healthy disk. Please visit my blog page and seach for "adding disk", there are some commands to run after adding disk. Also, please simulate that event, eg by using virutal machine or usb sticks to be sure. blog.datamaster.hr/8-myths-about-storage-spaces/
Storage Spaces Direct for sure do, but in plain Storage Spaces - I have never tried to be honest. For frequently accessed files that might give some benefit, but for other files not at all. I prefer pushing the limits, eg. having as many as you can M.2 (or plain SATA) drives and stripe on that creating Virtual Disk with maximum column count to give you max performance (very close to single disk speed x num. of disks). Creating another virtual disk with redundancy (eg. mirroring) on the same storage pool of disks for important files is wise, plus a backup somewhere else of course.
@@VedranKesegicSQL *Well, you'll love this video: How Fast Is Motherboard RAID with Five SSD's? ua-cam.com/video/Rv8ouhYvJ14/v-deo.html Although his video is based-on DMI2 (DMI3 is currently available), his points can be easily applied to DMI3. Just a few points: *At-best stripping speeds only works-well for large sequential files (ex: photo or video editing) & not very well for small or random file. *Note striping benefits require independent/parallel connections (aka parallel lanes) to each physical Disk. Although PCIe & SATA3 ports have parallel lanes to the cpu, USB3 ports don't. Windows10 doesn't offer SSD caching but intel cpus does: www.lifewire.com/intel-smart-response-technology-833444 Tell me what you think
Thanks Mihai! I needed Storage Spaces as a good way to drastically improve performance of SQL Server databases in cloud environments. So this video was born to share my excitement with technology.
3 роки тому+1
@@VedranKesegicSQL nice! I also have to deal with databases in the cloud, as part of a current university course, so this 'trick' is pretty handy! hope that you don't mind if I 'steal' it a bit, hahah! :) anyway, thank you for all the valuable info!
When you say that 2 drives can be mirrored and striped... they aren't really *BOTH* mirroring and stripping your data. (RAID 10) They are just "this data is mirrored" and "this other data is being stripped". Very different than "mirroring and striping my data".
Exactly! In drawing demo, we have two physical disks in the storage pool. On top of them we built one virtual disk (VD) mirrored, and one virtual disk striped. Both VD with different resiliency over same "shared" two physical disks.
Vedran, I just got a 4bay HDD USB3.1Gen2 enclosure ( OWC Mercury Elite Pro Quad) & am running 3x8TB mybook HDDs in parity under Storage Spaces. I'm still trying to figure-out how to use my extra 2TB SSD to cache the three HDDs though? SSD Caching the HDDs would decrease their workload, increase data access performance, etc. It seems my least worst option is PrimoCache, Thks again.
I tried disk caching software on a PC, but that simply didn't turn out well with my PC daily restarts. Caching inside of a dedicated device (NAS) works better IMHO. Storage Spaces do have "tiering" which achieves similar goal by placing often used data to faster disks (SSD). I have never tried it myself though. You will probably have to use powershell, examples are here: www.miru.ch/creating-tiered-storage-spaces-in-server-2012-r2/ For a plug and play solution, you would probably buy a NAS and put your drives inside, something like Synology DS1019+ or DS918+ www.synology.com/en-global/products/DS1019+
@@VedranKesegicSQL Thks, I watched a few powershell introduction videos & my head is still spinning. So I hope to eventually find a powershell geek & get him to setup SSD caching up for me. Microsoft will probably put SSD caching into the Windows11 anyways. Oh it's Synology's !DS720+ or Burst! Speeding up reading files on a Synology NAS (local caching of NAS files)ua-cam.com/video/eryytxpHR0A/v-deo.html Here's a good Storage Spaces video: $4,000 Ultimate Computer Build - Part 8 - Storage Spaces ua-cam.com/video/15XP2WqyFIw/v-deo.html Computer Backup - Are You Prepared? - Guide to Safe & Easy Backups ua-cam.com/video/319Tn7w2rM4/v-deo.html
@@VedranKesegicSQL Vedran, It seems like Synology's drive offers the most stable SSD caching of USB HDDs especially over a network ; Speeding up reading files on a Synology NAS (local caching of NAS files) ua-cam.com/video/eryytxpHR0A/v-deo.html . So I'll just plug my USB HDD enclosure into a DS720+ (when it becomes available) & use Synology's drive as a go between (at least according to their manual). Unfortunately Synology won't let you apply striping/mirroring/parity to non-Synology attached storage and the cost$ per NAS HDD bay goes up exponential. Hopefully the next version of Windows OS will include stable SSD caching of USB HDDs.
Great video, thanks for the info! I still have one question. If I have two mirrored drives and take one out - can the data of this drive be read on another pc if I put the drive in there?
I think yes, in theory, it should be readable on other PC. And if you add an empty drive there, you should be able to rebuild the mirror on that new PC. Please test and write your experience here.
Very good info! One thing I’m confused by is I have two 6tb drives mirrored and want to expand this storage pool. I am looking at getting two 18tb drives. Can I add these two 18tb drives to the same pool of 6tb? I assume I cannot since the two 6tb drives will not be able to mirror anything past their capacity. Essentially I will have to start a separate storage pool for 18tb right? Basically the question is for mirrored storage pools do drives added on need to be the same size? Also if it is mirrored I would need to add two more drives. Adding only one more drive is only possible for parity right?
Hi. To existing 2x6TB you can add 2x18TB to the same pool and have capacity expanded with 0 wasted space (all space is usable). You can even add 1x18TB disk, but then part of space would be not usable. Storage spaces is very smart and efficient in achieving max possible capacity with any drives you throw at it. Exactly how much capacity you get - use my Storage Spaces Capacity Calculator here. blog.datamaster.hr/capacity-calculator-for-storage-spaces/ For 6,6,18,18 you get 24TB with no wasted space. For 6,6,18 you get 12TB capacity with 6TB wasted (unusable). Minus 10% because of 1024 vs 1000 difference, that is 24TB = 21.6TB, 12TB = 10.8TB of capacity. All assuming 2-way Mirroring with Column Number = 1.
@@VedranKesegicSQLohhh I get what you are saying. So it’s only two way redundancy no matter what I add it will just be divided by two and rebalanced between newly added drives. So [6+18 | 6+18]
Nice Video! For the speed test, it would be nicer to test it with real independent drives. So start with two drives and measure it, and then add a third drive (all the same model) and measure again. Adding just a faster SSD does not prove that adding a drive increases performance, because it would have had the same effect if you just replaced one drive.
Hello Vedran , i have a question . I have a storage pool with 3 500gb drives. I want to upgrade with 3 1Tb drives , is there any tip how to know how much is the new usable space i will have. I read somewhere that for this you make a sumary of all drives capacity , lets say is 3000gb so 3000*66.7% = 2001gb and thats the new usable capacity...is this correct???
Hi! You are like reading my mind, I just thought to make a video around usable space from mixed drives. It depends on number of columns of your virtual disk, and type (simple, 2-way mirror, parity, ...). Can you give that info? Get-VirtualDisk should display that.
@@arxigos21 If you use resiliency "simple" (no resiliency) and columns=3 then capacity "before" is 1,5TB and after 4,5TB. If resiliency is PARITY and columns=3, capacity "before" is 1TB and after 3TB. Some space goes for metadata for expand operation, cca 1GB. In the video there will be a calculator you can use to predict capacity with any combination of disks, any resiliency, and it will tell how much space (if any) will be left unused becuase of uneven disk sizes.
@@arxigos21 Blog is finished, video recorded but not yet ready. You can use logic and the TSQL stored procedure code from there: blog.sqlxdetails.com/capacity-calculator-for-storage-spaces/
What I never understand about this storage space thing nobody ever mentioned if the operating system drive can also be part of the storage space pool. Even when you google this, there will be no information on this.
Hi. Thank for this very good question. Seems it is not supported, unless particular UEFI firmware you have has that capability, as some do: superuser.com/questions/1400960/is-it-possible-to-boot-windows-1809-system-from-storage-spaces
Hello i currently have an 8TB that is completely full however I'm adding another 8tb to make it 16tb. The question is will i lose data on the original 8tb if i create a simple (no resiliency) drive?
No, you won't lose data. Long answer: if you already have created an 8TB Virtual Disk (Storage Space) on a pool that contained only 1 8TB HDD, by adding one more 8TB HDD to the pool, you will be able to expand your virtual disk to 16TB (a bit less) without losing any data. It doesn't matter if it is "simple". Redundancy only means that data can survive a one disk failure (or two), while "simple" will not survive a disk failure (data is lost if a drive fails).
Hi, I've tried to figure out if creating a storage space in Single parity (similar to RAID 5) setup but using only external drives connected to an older Intel NUC with Win10. If I were to upgrade that NUC in 1-2 years still using Win10, would there be any issues with that process. Ie. simply connect the external harddrives to the new NUC and power it all on afterwards? Plan to run a small PLEX server on that setup. I've tried searching here and there, but it might be my limited vocabulary to why I cannot find some answers to this setup :D
It should work out of the box. Metadata about virtual disks is stored on every physical disk. When you plug them to a new machine, windows knows which disks should be there, entire Storage Spaces configuration is there, and virtual disks should appear on that new machine. In theory at leat. Please post here if it succeeds or fails, either experience is appreciated to know!
Vedran, I finally found someone who figured-out how to SSD cache his HDDs within Windows10 using Powershell. In-theory now all I have to do is focus on his ~10 lines of Powershell cmds. I'm still trying to find a Powershell expert to set it up for me though. It's ashamed no Powershell guru haven't posted a UA-cam video on exactly how non-geek can do this. Yeap, I make-up for not being smart with determination & persistence (aka being stubborn ;). Thks for the help along the way. Tiered Storage Spaces Windows 10 linustechtips.com/main/topic/925425-tiered-storage-spaces-windows-10/ "......First I use the Storage Spaces GUI to create a new pool and storage space, selecting all 4 2TB standard SATA hdd's (3x 2TB WD Green Cavier's at 3Gb/s 5400 RPM and 1x 2TB WD Red 6Gb/s so it drops to 3 GB/s to match the other 3 and I include the 1 single 120 GB SSD which is a Intel® SSD 520 Series. In the GUI I'm forced to create the Storage space for the pool, so I just accept my defaults and click create storage space. I then delete that storage space, keep the storage pool but rename it to "mypool". I then ran these commands in powershell: Get-StoragePool mypool | Set-ResiliencySetting -Name Simple -NumberOfColumnsDefault 1 Get-StoragePool mypool | Set-ResiliencySetting -Name Mirror -NumberOfColumnsDefault 2........"
Thanks. Your video is very informative. I still have a way to go to solve my particular situation. Don't know how it happened but for some reason, my main system SSD is showing as accompanying the 4 physical SSDs inside an SSD enclosure, to make up a Storage Pool. That system SSD shows as having an "Error." The Storage Pool should only include the 4 SSDs. To add to the confusion, in the Physical Drives section of Storage Spaces, one of the 4 SSDs that occupy the enclosure, shows an error, although in the Device Manager, all 4 SSDs show "working properly." I am very apprehensive to remove the system SSD from the Storage Pool considering what might happen with its data.
@@VedranKesegicSQL Thanks Vedran for such a speedy reply...as I mentioned, I don't know how it happened..must have been something I did when I was using Macrium Reflect to clone an undersized system disk to the one I have now. I will take your advice and backup before removing it from the Pool. Thanks again!
Yes. Adding a drive to the storage pool erases data from that drive. So you need to copy the data elsewhere before you join the drive to the pool, create a virtual disk, and then copy data back.
Great video and detail. I was a little confused when you have RAID 1 and 0. Was this configuration like 1 + 0? Because, I was unsure when you pulled a drive from the RAID 0 Stripe, you went to the RAID 1 Mirror and made a new file? Isn't it about what data is on each set? Thanks
It is not raid1+0. Only 2 physical disks were used in the storage pool, but on them are created two virtual disks, thin provisioned, one with mirroring (raid 1) and one with striping (raid 0). Very flexible.
@@VedranKesegicSQL So, could the data on 1 set could be replicated to the other? I'm trying to find a setup that would work for me with redundancy of all files yet have some perf.
@@timetorelaxfocus9642 Virtual Disk which was created as mirrored, automatically copis data to two physical disks and can survive one disk failure as shown in the video. Mirroring (2-way mirror) has better performance than parity (raid5), so go with that. Plus backups of course.
I have never seen such a thing. Can you give a little more details (what is "2004 update", VD configuration, did that happened to you or someone else, etc)? I hope this is not trolling.
Hi Rifat. What do you compare NVMe setup with? What is the NVMe setup you talk about? Assuming you setup NVMe into Storage Pool and setup Virtual Disks, the reason for slowness could be that you didn't specify "Column Number" while creating new Virtual Disk (possible only though powershell and only in VD creation moment, NOT changeable later!). If you ommit to set column count or you set it low, or you just click "next, next, finish" in wizard, total speed is usually the speed of the slowest disk. There are NVMe disks that are garbage in write speed. They start fast (eg 3500 MB/s) and after 2 seconds they drop to eg 50 MB/s because they exhaused the write cache. That is the case with cheap NVMe disks. Search for "sustained write test" when evaluating a NVMe to buy. Also, if you use "Parity" (aka RAID5) with Storage Spaces, that us super-slow. Instead, use Mirroring or Simple (no resiliency) instead. Also upgrade and patch windows to the newest, and check SS version (upgrade pool and VD if needed to a newer version of starage spaces by using command "Update-StoragePool"). There is possibility of imbalance of data across disks too, which can be corrected by the powershell command “Optimize-StoragePool”.
@@RifatErdemSahin So far no video, but here is example (I would use thin provisioning, simple/no resiliency and column number = num of disks = 4 - then test the speed!): dannyda.com/2020/01/03/use-powershell-to-create-virtual-disk-on-windows-storage-pool-storage-space-with-specific-number-of-columns/
Sorry but the presentation lacks one important thing that is also omited on most videos about storage spaces. How does work stripped on more than two disks? Will it spread the same file across all disks, or will it put one file on one disk, other file on different disk, etc. Main idea, how would write speed will be affected by adding more disks. And the other part, how does two way resiliant works on three drives? Since it puts only two copies, how does it decide on what disk each copy is written and would write speed be increased if added more disks (i doub it). Note: Case use would be having only one big file of more than two hundred gigabytes, just to ensure it would work as spected, at block level, and not at file level. Extra: Some filesystems do similar to raid0 by putting some files on one disk (full file is only on one disk) and other files on another drive, but each file is stored completly on the same disk, files are not cutted, not striped, so if you loose one drive youonly loose files stored onthat drive, the rest can be read/write without problem; and for similar to raid1 (only two copies) it does one copy on one disk and another copy onanother disk, balancing free space on all N disks. That is way i ask how it works, beacuse there are a lot of waya to do itRAID0, RAID1, RAID10, RAID5 (block level), and at per file level BTRFS raid0, BTRFS raid1, BTRFS raid10 also with different disks sizes, etc.
Hi. SS splits data in blocks (called stripes I think), not files. How many disks in parallel will be used is determined by parameter called "column count". If number of columns is 2 in striping (no resiliency), that means 2 disks will bi used at once on every access. If you have 3 disks, column count 2, and you write one big file there, 2 disks will receive one piece of data and third wont. Then it shifts the drive that wont receive the data so all disks are evenly filled. Eg, if drives are A,B,C it writes first two chunks to AB then AC, then BC, then all over again until the file write is completed.
@@VedranKesegicSQL Thanks, so it looks loke by block (stripe) not by full file, more siilar to RAID (in uppercase) than BTRFS raid (in lowwercase). For examploe in BTRFS raid0 with three disks, you disconect one drove (or two drivers at the same time)and some files can be readed completly, not all data is lost, since each file is stored only once but all of it is stored on the same drive. This is what i think based on my own tests with BTRFS on six SSD, when i had tested BTRFS (raid0, raid1 and raid10)... and i had also tested on BTRFS theese steps 1 Use two disks 2. Write some files 3. Unplug one drive without telling nothing to BTRFS while system is offline 4. Write more files with one missing drive 5. Unplug the other drive without telling nothing to BTRFS while system is offline 6. Replug the first unplugged drive without telling nothing to BTRFS while system is offline 7. Write more files with one missing drive 8. Replug the first unplugged drive without telling nothing to BTRFS while system is offline 9. Now depending on what drive i use for mount command i can see files written on step 4 or on step 7 and save them on a different drive (third drive) 10. Do a re-build (BTRFS balance) and put back that files that are now missing 11.- No data lost So the raid must be done at file level, each file must not be saved on both drives (part of it on one drive, part on another), since doing so, when one disk is not plugged such file would not be readable, but it was. I had not see any other kind of RAID0 like that, all the rest if one drive is missing you loose all. The bad side of BTRFS raid0 does not improve write & read speeds if not asked to do it in different files, at least is what i had seen. That different way of 'raid' is what make me ask how really works Windows Storage Spaces at low level... since i am planing to use three SSD on an old PCIe Gen 2 16x (motherboard port) with a non hardware raid PCIe Gen 3 x4 SATA III controller, motherboard SATA has only two ports and both only SATA II... and for big files (virtual machine disks in inmutable state with differencing file in non auto reset) each between 10 to 30 gigabytes. Thanks for confirming that Windows Storage Spaces works at block level (stripes) ad not per file, now the next question (mostly for performance reasons) what would be the size of such stripes (128KiB, 1MiB, configurable or not, etc) and i must investigate about ReFS instead on NTFS also, any tip on both is wellcome; also if it would be OK to use such for ofimatic files, etc. and a last point, stripe size is used as save unit on each drive or is the block size that is divided onto all disks (for Raid0)... if 4 disks and stripe is 128KiB would it write 128KiB on each drive or 128KiB/4 on each drive. Different raid implementations do it differently... some hardware raid0 controllers requiere 2^N identical drives, others canuse 3, 5, 7 and any N>=2 number of drives. By the way, i had seen that drives must be >=4GiB to be able to be used on Windows Storage Spaces, but on Windows screen it does not say such thing, or at least i did not find where... my 3.82GiB (4GB) USB mass storage did not show on Storage Spaces but when i create a virtual disk of 4GiB it is shown... so that was my guess, at least 4GiB. Thanks in advance.
@@androidlogin3065 Correct terminology is: "chunk" is block written to one disk, also called "Interleave" - default is 256KB. Stripe is all chunks of multiple disks written or read together (at once). You can set these at creation time of Virtual Disk. Details are in this video: ua-cam.com/video/it-1WtNrNr4/v-deo.html
@@VedranKesegicSQL Thanks a lot, i am not an expert and want to know. 256KiB for chunk size looks like a lot if i undestand it correctly thay 256KiB chunk for four drives in raid0 equivalent would be 1MiB per read / write operation. I had added a comment on that video you suggest (after seeing it fully), since after seeing it i am more confused than before.
YOu are in for a rude awakening looking for speed increases from adding a drive in parity with actual SATA drives....; performance plummets from 110-140 M/sec sequential writes down to 32 MB/sec.... Mirrored? 120-130 MB/sec! But with 3 drives w/ parity? 30 MB/sec...! Yayyyyyyyy!!!!!!!!!
I love your videos, but when you make a list of "these are all myths" they shouldn't be just QUESTIONS. They should be "these are false statements, and I'll explain why". A "question" can't be true nor false.... it's just a question.
Because? If you refer to bugs it had N years ago, that was fixed. I have seen many production system using Storage Spaces with great success. No problems whatsoever.
Great video. Deserves more views.
Thanks R5! Please spread the word / link...
Good work and explanations. Storage Spaces is Microsofts hidden gem.
Great tutorial! My question: My homeserver for backup has 7 drives in the storage space. Total 7 TB. If my motherboard goes kaput, will the new system configure the pool by itself? Or do I need to backup the backup server?
Short answer: no, you do not have to backup the server. Long answer: Configuration of the pool and virtual disks is stored onto disks themselves. That is why "metadata" shaves a little bit off of capacity. If you remove all the drives, and plug them into a new Windows system that supports Storage Spaces, it will recognize configuration of virtual disks and should automagically appear there without any intervention from you, except plugging the drives in.
I've been using Storage Spaces for 8-9 years on a Windows 8 to 10 computer that is being used a server. I have a two-way mirror and simple space set up. In all of that time I've never had a single issue with it. My only gripe is the inability to reduce thin provisioning size and the inability to make non-thin provisioned volumes. Minor quibbles.
Can someone list a few real-world reasons for "show that I have a drive that's much bigger than I really have"? Other than being EXTREMELY misleading... what's the point? The extra space can never be used. Why not keep all the numbers accurate, and then when you really add the new disk *THEN* it will show the extra available space. Why would I want to falsely believe that I have a 4TB drive that is only 50% full, when it's really just a 2TB drive that's 99.99% full? Dangerous.
12:07 The question you ask yourself after using Windows storage space
(Edit: too bad current lastest update can't format properly, so I created the storage pool and attempted to create a drive. Its online in disk management, so I delete the volume and format it from there. I don't know if that's a legit way.)
After watch your video a few times, I will keep my:
Least critical files on the stripping w/o parity drive
Middle critical files on the stripping with parity drive
Most critical files on the mirroring drive
all across three or more physical disks.
Thks
Hello. Any pointers, tips, tutorials on how to speed the pool up using SSDs as a tier? Thanks.
Thanks for an idea. Unfortunately, I have not yet used tiers in Storage Spaces. To achieve high performance, I used multiple columns (configured high "column count" number) when creating virtual disks. That makes disks run in parallel, much faster than single drive. See here: ua-cam.com/video/rl7t8FToDsU/v-deo.html
@@VedranKesegicSQL Thanks for the link!
I found this wiki, but not sure why the sizes are marked like they are.
wiki.towaso.de/index.php/Create_Tiered_Storage_Space
Also found this, which is what I'm going to try with 2 x SSDs and 3 x HDDs:
forum.level1techs.com/t/how-to-set-up-tiered-storage-on-windows-10/126840/20
Thks ever so much, I just ordered a few USB thumbdrives to replicate your 'Storage Spaces' video.
If it works half as well, I'll be able to backup my ~40GB data on multiple mirrored thumdrives, keep one in the car, & once ever few days sync the one in the car to the other mirrored thumdrives.
Oh, you didn't try reconnecting the failed disk. If you ever update the video, you might try that.
Huh, USB drives were just for demo purpose to be able to pull them out to simulate failure. For actual backup you would be probably better with some cloud provider (eg. google drive, mega.nz, dropbox, onedrive). For bigger sets of data, a NAS is a decent option, even the smallest one like described here: blog.sqlxdetails.com/crypto-virus-resistent-backup/
Storage Spaces are not intended for backup purpose, just for resiliency to disk failure and speed, sort of more flexible RAID. It is not intended to work with one drive "unplugged" for a long periods of time. Might work though, but that is not what it is made for.
@@VedranKesegicSQL Roger-that & thks for the concern,
*I have a Z2 Mini G4 right. Also I'll probably get a OWC Mercury Elite Pro Quad & manage it via MSWindow Storage Spaces. Oh, I saving my thunderbolt3 port for an external GPU.
HP Z2 Mini workstation PC ua-cam.com/video/qcjNm9i_pws/v-deo.html
OWC Mercury Elite Pro Quad hdd bay enclosure (10Gb/s USB 3.1 Gen 2) eshop.macsales.com/item/OWC/MEQCTJBT00/
*Over 6 months ago I was focused on a Synology NAS but their 4&above HDD bay NASes cost an arm&leg as well as contain weak CPUs. Now I plan to keep my mass-storage attached to my workstation & get an inexpensive 2bay Synology NAS mainly for interfacing with/over the internet from my home.
Synology DiskStation NAS Introduction ua-cam.com/video/CaInLiK4zFM/v-deo.html
Synology - Play. Collaborate. Achieve. Live BeyondCloud ua-cam.com/video/UsnBqL_pzcs/v-deo.html
Synology - Global Event Introduction Video ua-cam.com/video/IXPnP__Ipvc/v-deo.html
Synology - Be Your Own Cloud ua-cam.com/video/QOtW9CBsTys/v-deo.html
Synology | Introduction to Synology Collaboration Suite ua-cam.com/video/M9OZ-i9N8NA/v-deo.html
Eventually I'll emulate everything the Synology NAS does for me in Linux on my workstation but that's take a couple of years (I'm a bored retired physicist). Of course *I'll do versioned/near-real-time backups. But I currently have only ~20GB of personnel data I really care about. It would be nice to mirror/ecrypt it to three thumbdrives, normally keep one thumbdrive in my vehicle, & resync that one in my car once or twice a week. When a thumbdrive dies, I just destroy it & get another one. If Storage Spaces won't do it, I try some other way.
*I like the simplicity of Storage Spaces though.
Let me know what you think.
This is great illustration video thank you, i would like to add that moving storage disks into another system will keep the storage spaces configuration same ... works like a charm.
I have been using storage spaces for almost 4 years and i totally in love with it, i custom build my home HTPC with hostswap drives and using storage spaces as media backup and network file-share storage. its a grate solution if you don't want to lock your data with hardware raid motherboards/cards or invest in NAS.
SS is certainly great for a simple 2-drive mirror...; 1 drive crashes/fails, nothing lost at all. (Unless you are OK with very slow writes, I've avoid parity raids with storage spaces, as the write speeds are abysmal!)
Great video man. Thank you for all that information!
I have one question please. At the last of the video you remove the USB like a hdd failure. After a while lot of warnings in storage space management. But what we can do after. Let's say that this usb was a hdd that suddenly died! How can we add a new replacement and tell to the storage Spaces to start recreating the data in the new drive? We can't let them like that and work we the warnings right? I have found that we can first remove and the add a drive but I don't know how we can do that when some of our drives is died. No option for remove imas it seems in your video.
Thank you man again
If a disk dies, you can add a new healthy disk. Please visit my blog page and seach for "adding disk", there are some commands to run after adding disk. Also, please simulate that event, eg by using virutal machine or usb sticks to be sure. blog.datamaster.hr/8-myths-about-storage-spaces/
@@VedranKesegicSQL Thank you! I will check that.
Brilliant tutorial, thank you so much for making and sharing it. Subscribed!
Thanks Mick for nice words!
Well Done!, Thanks for clearing up the terminology. I will give it a try.
Question: ?Can Windows 10 Pro provide HDD caching on a SATA3 SSD?
Storage Spaces Direct for sure do, but in plain Storage Spaces - I have never tried to be honest. For frequently accessed files that might give some benefit, but for other files not at all. I prefer pushing the limits, eg. having as many as you can M.2 (or plain SATA) drives and stripe on that creating Virtual Disk with maximum column count to give you max performance (very close to single disk speed x num. of disks). Creating another virtual disk with redundancy (eg. mirroring) on the same storage pool of disks for important files is wise, plus a backup somewhere else of course.
@@VedranKesegicSQL *Well, you'll love this video:
How Fast Is Motherboard RAID with Five SSD's? ua-cam.com/video/Rv8ouhYvJ14/v-deo.html
Although his video is based-on DMI2 (DMI3 is currently available), his points can be easily applied to DMI3.
Just a few points:
*At-best stripping speeds only works-well for large sequential files (ex: photo or video editing) & not very well for small or random file.
*Note striping benefits require independent/parallel connections (aka parallel lanes) to each physical Disk.
Although PCIe & SATA3 ports have parallel lanes to the cpu, USB3 ports don't.
Windows10 doesn't offer SSD caching but intel cpus does: www.lifewire.com/intel-smart-response-technology-833444
Tell me what you think
@Casix03 Thks again
great video! just stumbled upon this feature of Windows and found your video on YT regarding this. cool presentation!
Thanks Mihai! I needed Storage Spaces as a good way to drastically improve performance of SQL Server databases in cloud environments. So this video was born to share my excitement with technology.
@@VedranKesegicSQL nice! I also have to deal with databases in the cloud, as part of a current university course, so this 'trick' is pretty handy! hope that you don't mind if I 'steal' it a bit, hahah! :) anyway, thank you for all the valuable info!
@ Sure, I'm glad it is useful ;)
When you say that 2 drives can be mirrored and striped... they aren't really *BOTH* mirroring and stripping your data. (RAID 10) They are just "this data is mirrored" and "this other data is being stripped". Very different than "mirroring and striping my data".
Exactly! In drawing demo, we have two physical disks in the storage pool. On top of them we built one virtual disk (VD) mirrored, and one virtual disk striped. Both VD with different resiliency over same "shared" two physical disks.
Vedran, I just got a 4bay HDD USB3.1Gen2 enclosure ( OWC Mercury Elite Pro Quad) & am running 3x8TB mybook HDDs in parity under Storage Spaces. I'm still trying to figure-out how to use my extra 2TB SSD to cache the three HDDs though? SSD Caching the HDDs would decrease their workload, increase data access performance, etc. It seems my least worst option is PrimoCache, Thks again.
I tried disk caching software on a PC, but that simply didn't turn out well with my PC daily restarts. Caching inside of a dedicated device (NAS) works better IMHO. Storage Spaces do have "tiering" which achieves similar goal by placing often used data to faster disks (SSD). I have never tried it myself though. You will probably have to use powershell, examples are here: www.miru.ch/creating-tiered-storage-spaces-in-server-2012-r2/
For a plug and play solution, you would probably buy a NAS and put your drives inside, something like Synology DS1019+ or DS918+ www.synology.com/en-global/products/DS1019+
@@VedranKesegicSQL Thks, I watched a few powershell introduction videos & my head is still spinning. So I hope to eventually find a powershell geek & get him to setup SSD caching up for me. Microsoft will probably put SSD caching into the Windows11 anyways.
Oh it's Synology's !DS720+ or Burst!
Speeding up reading files on a Synology NAS (local caching of NAS files)ua-cam.com/video/eryytxpHR0A/v-deo.html
Here's a good Storage Spaces video:
$4,000 Ultimate Computer Build - Part 8 - Storage Spaces ua-cam.com/video/15XP2WqyFIw/v-deo.html
Computer Backup - Are You Prepared? - Guide to Safe & Easy Backups ua-cam.com/video/319Tn7w2rM4/v-deo.html
@@VedranKesegicSQL Vedran, It seems like Synology's drive offers the most stable SSD caching of USB HDDs especially over a network ; Speeding up reading files on a Synology NAS (local caching of NAS files) ua-cam.com/video/eryytxpHR0A/v-deo.html
.
So I'll just plug my USB HDD enclosure into a DS720+ (when it becomes available) & use Synology's drive as a go between (at least according to their manual). Unfortunately Synology won't let you apply striping/mirroring/parity to non-Synology attached storage and the cost$ per NAS HDD bay goes up exponential.
Hopefully the next version of Windows OS will include stable SSD caching of USB HDDs.
Great video, thanks for the info!
I still have one question. If I have two mirrored drives and take one out - can the data of this drive be read on another pc if I put the drive in there?
I think yes, in theory, it should be readable on other PC. And if you add an empty drive there, you should be able to rebuild the mirror on that new PC. Please test and write your experience here.
I get conflicting reports on the speed
Very good info! One thing I’m confused by is I have two 6tb drives mirrored and want to expand this storage pool. I am looking at getting two 18tb drives. Can I add these two 18tb drives to the same pool of 6tb? I assume I cannot since the two 6tb drives will not be able to mirror anything past their capacity. Essentially I will have to start a separate storage pool for 18tb right? Basically the question is for mirrored storage pools do drives added on need to be the same size? Also if it is mirrored I would need to add two more drives. Adding only one more drive is only possible for parity right?
Hi. To existing 2x6TB you can add 2x18TB to the same pool and have capacity expanded with 0 wasted space (all space is usable). You can even add 1x18TB disk, but then part of space would be not usable. Storage spaces is very smart and efficient in achieving max possible capacity with any drives you throw at it. Exactly how much capacity you get - use my Storage Spaces Capacity Calculator here. blog.datamaster.hr/capacity-calculator-for-storage-spaces/
For 6,6,18,18 you get 24TB with no wasted space. For 6,6,18 you get 12TB capacity with 6TB wasted (unusable). Minus 10% because of 1024 vs 1000 difference, that is 24TB = 21.6TB, 12TB = 10.8TB of capacity. All assuming 2-way Mirroring with Column Number = 1.
@@VedranKesegicSQLohhh I get what you are saying. So it’s only two way redundancy no matter what I add it will just be divided by two and rebalanced between newly added drives. So [6+18 | 6+18]
@@RED89P13 It can be more complicated than that, but in a nutshell - yes.
Nice Video! For the speed test, it would be nicer to test it with real independent drives. So start with two drives and measure it, and then add a third drive (all the same model) and measure again.
Adding just a faster SSD does not prove that adding a drive increases performance, because it would have had the same effect if you just replaced one drive.
Hello Vedran , i have a question . I have a storage pool with 3 500gb drives. I want to upgrade with 3 1Tb drives , is there any tip how to know how much is the new usable space i will have. I read somewhere that for this you make a sumary of all drives capacity , lets say is 3000gb so 3000*66.7% = 2001gb and thats the new usable capacity...is this correct???
Hi! You are like reading my mind, I just thought to make a video around usable space from mixed drives. It depends on number of columns of your virtual disk, and type (simple, 2-way mirror, parity, ...). Can you give that info? Get-VirtualDisk should display that.
@@VedranKesegicSQL ok thanks i will wait for this video :)
@@arxigos21 If you use resiliency "simple" (no resiliency) and columns=3 then capacity "before" is 1,5TB and after 4,5TB. If resiliency is PARITY and columns=3, capacity "before" is 1TB and after 3TB. Some space goes for metadata for expand operation, cca 1GB. In the video there will be a calculator you can use to predict capacity with any combination of disks, any resiliency, and it will tell how much space (if any) will be left unused becuase of uneven disk sizes.
@@VedranKesegicSQL Yes , thats exactly whan i need , thanks a lot. subscribed...
@@arxigos21 Blog is finished, video recorded but not yet ready. You can use logic and the TSQL stored procedure code from there: blog.sqlxdetails.com/capacity-calculator-for-storage-spaces/
What I never understand about this storage space thing nobody ever mentioned if the operating system drive can also be part of the storage space pool. Even when you google this, there will be no information on this.
Hi. Thank for this very good question. Seems it is not supported, unless particular UEFI firmware you have has that capability, as some do: superuser.com/questions/1400960/is-it-possible-to-boot-windows-1809-system-from-storage-spaces
Hello i currently have an 8TB that is completely full however I'm adding another 8tb to make it 16tb. The question is will i lose data on the original 8tb if i create a simple (no resiliency) drive?
No, you won't lose data. Long answer: if you already have created an 8TB Virtual Disk (Storage Space) on a pool that contained only 1 8TB HDD, by adding one more 8TB HDD to the pool, you will be able to expand your virtual disk to 16TB (a bit less) without losing any data. It doesn't matter if it is "simple". Redundancy only means that data can survive a one disk failure (or two), while "simple" will not survive a disk failure (data is lost if a drive fails).
Awesome video :)
Thanks Mattie!
Hi, I've tried to figure out if creating a storage space in Single parity (similar to RAID 5) setup but using only external drives connected to an older Intel NUC with Win10. If I were to upgrade that NUC in 1-2 years still using Win10, would there be any issues with that process. Ie. simply connect the external harddrives to the new NUC and power it all on afterwards? Plan to run a small PLEX server on that setup. I've tried searching here and there, but it might be my limited vocabulary to why I cannot find some answers to this setup :D
It should work out of the box. Metadata about virtual disks is stored on every physical disk. When you plug them to a new machine, windows knows which disks should be there, entire Storage Spaces configuration is there, and virtual disks should appear on that new machine. In theory at leat. Please post here if it succeeds or fails, either experience is appreciated to know!
Vedran, I finally found someone who figured-out how to SSD cache his HDDs within Windows10 using Powershell. In-theory now all I have to do is focus on his ~10 lines of Powershell cmds. I'm still trying to find a Powershell expert to set it up for me though. It's ashamed no Powershell guru haven't posted a UA-cam video on exactly how non-geek can do this.
Yeap, I make-up for not being smart with determination & persistence (aka being stubborn ;). Thks for the help along the way.
Tiered Storage Spaces Windows 10 linustechtips.com/main/topic/925425-tiered-storage-spaces-windows-10/
"......First I use the Storage Spaces GUI to create a new pool and storage space, selecting all 4 2TB standard SATA hdd's (3x 2TB WD Green Cavier's at 3Gb/s 5400 RPM and 1x 2TB WD Red 6Gb/s so it drops to 3 GB/s to match the other 3 and I include the 1 single 120 GB SSD which is a Intel® SSD 520 Series. In the GUI I'm forced to create the Storage space for the pool, so I just accept my defaults and click create storage space. I then delete that storage space, keep the storage pool but rename it to "mypool".
I then ran these commands in powershell:
Get-StoragePool mypool | Set-ResiliencySetting -Name Simple -NumberOfColumnsDefault 1
Get-StoragePool mypool | Set-ResiliencySetting -Name Mirror -NumberOfColumnsDefault 2........"
Thanks. Your video is very informative. I still have a way to go to solve my particular situation. Don't know how it happened but for some reason, my main system SSD is showing as accompanying the 4 physical SSDs inside an SSD enclosure, to make up a Storage Pool. That system SSD shows as having an "Error." The Storage Pool should only include the 4 SSDs. To add to the confusion, in the Physical Drives section of Storage Spaces, one of the 4 SSDs that occupy the enclosure, shows an error, although in the Device Manager, all 4 SSDs show "working properly." I am very apprehensive to remove the system SSD from the Storage Pool considering what might happen with its data.
You should not be able to add a boot disk to the storage pool, it is not "eligible" for that. Do a backup before trying to fix it.
@@VedranKesegicSQL Thanks Vedran for such a speedy reply...as I mentioned, I don't know how it happened..must have been something I did when I was using Macrium Reflect to clone an undersized system disk to the one I have now. I will take your advice and backup before removing it from the Pool. Thanks again!
If you already have data on your drives, will creating a storage pool cause you to loose your data?
Yes. Adding a drive to the storage pool erases data from that drive. So you need to copy the data elsewhere before you join the drive to the pool, create a virtual disk, and then copy data back.
Thanks
Excellent video
Great video and detail.
I was a little confused when you have RAID 1 and 0. Was this configuration like 1 + 0?
Because, I was unsure when you pulled a drive from the RAID 0 Stripe, you went to the RAID 1 Mirror and made a new file?
Isn't it about what data is on each set?
Thanks
It is not raid1+0. Only 2 physical disks were used in the storage pool, but on them are created two virtual disks, thin provisioned, one with mirroring (raid 1) and one with striping (raid 0). Very flexible.
@@VedranKesegicSQL So, could the data on 1 set could be replicated to the other? I'm trying to find a setup that would work for me with redundancy of all files yet have some perf.
@@timetorelaxfocus9642 Virtual Disk which was created as mirrored, automatically copis data to two physical disks and can survive one disk failure as shown in the video. Mirroring (2-way mirror) has better performance than parity (raid5), so go with that. Plus backups of course.
@@VedranKesegicSQL Ok, thanks much
great video!
Thanks!
Why is the storage space destroyed by 2004 windows update?
Data is mostly corrupted and lost!
I have never seen such a thing. Can you give a little more details (what is "2004 update", VD configuration, did that happened to you or someone else, etc)? I hope this is not trolling.
why is it slower mate? on nvme setup ?
Hi Rifat. What do you compare NVMe setup with? What is the NVMe setup you talk about? Assuming you setup NVMe into Storage Pool and setup Virtual Disks, the reason for slowness could be that you didn't specify "Column Number" while creating new Virtual Disk (possible only though powershell and only in VD creation moment, NOT changeable later!). If you ommit to set column count or you set it low, or you just click "next, next, finish" in wizard, total speed is usually the speed of the slowest disk. There are NVMe disks that are garbage in write speed. They start fast (eg 3500 MB/s) and after 2 seconds they drop to eg 50 MB/s because they exhaused the write cache. That is the case with cheap NVMe disks. Search for "sustained write test" when evaluating a NVMe to buy. Also, if you use "Parity" (aka RAID5) with Storage Spaces, that us super-slow. Instead, use Mirroring or Simple (no resiliency) instead. Also upgrade and patch windows to the newest, and check SS version (upgrade pool and VD if needed to a newer version of starage spaces by using command "Update-StoragePool"). There is possibility of imbalance of data across disks too, which can be corrected by the powershell command “Optimize-StoragePool”.
@@VedranKesegicSQL thanks mate for the detailed answer. i use 4 firecuda 1 tb. discmark test has not shown speed boost. i use w 11
@@VedranKesegicSQL so any video on how t set column number?
@@RifatErdemSahin So far no video, but here is example (I would use thin provisioning, simple/no resiliency and column number = num of disks = 4 - then test the speed!): dannyda.com/2020/01/03/use-powershell-to-create-virtual-disk-on-windows-storage-pool-storage-space-with-specific-number-of-columns/
There is no “prepare drive for removal“ in Windows 11?
I don't have Win11 so cannot check. But, powershell should work always.
I don't think usb discs are allowed on home edition, stablebit works better and is safer.
Sorry but the presentation lacks one important thing that is also omited on most videos about storage spaces.
How does work stripped on more than two disks? Will it spread the same file across all disks, or will it put one file on one disk, other file on different disk, etc. Main idea, how would write speed will be affected by adding more disks.
And the other part, how does two way resiliant works on three drives? Since it puts only two copies, how does it decide on what disk each copy is written and would write speed be increased if added more disks (i doub it).
Note: Case use would be having only one big file of more than two hundred gigabytes, just to ensure it would work as spected, at block level, and not at file level.
Extra: Some filesystems do similar to raid0 by putting some files on one disk (full file is only on one disk) and other files on another drive, but each file is stored completly on the same disk, files are not cutted, not striped, so if you loose one drive youonly loose files stored onthat drive, the rest can be read/write without problem; and for similar to raid1 (only two copies) it does one copy on one disk and another copy onanother disk, balancing free space on all N disks.
That is way i ask how it works, beacuse there are a lot of waya to do itRAID0, RAID1, RAID10, RAID5 (block level), and at per file level BTRFS raid0, BTRFS raid1, BTRFS raid10 also with different disks sizes, etc.
Hi. SS splits data in blocks (called stripes I think), not files. How many disks in parallel will be used is determined by parameter called "column count". If number of columns is 2 in striping (no resiliency), that means 2 disks will bi used at once on every access. If you have 3 disks, column count 2, and you write one big file there, 2 disks will receive one piece of data and third wont. Then it shifts the drive that wont receive the data so all disks are evenly filled. Eg, if drives are A,B,C it writes first two chunks to AB then AC, then BC, then all over again until the file write is completed.
@@VedranKesegicSQL Thanks, so it looks loke by block (stripe) not by full file, more siilar to RAID (in uppercase) than BTRFS raid (in lowwercase).
For examploe in BTRFS raid0 with three disks, you disconect one drove (or two drivers at the same time)and some files can be readed completly, not all data is lost, since each file is stored only once but all of it is stored on the same drive.
This is what i think based on my own tests with BTRFS on six SSD, when i had tested BTRFS (raid0, raid1 and raid10)... and i had also tested on BTRFS theese steps
1 Use two disks
2. Write some files
3. Unplug one drive without telling nothing to BTRFS while system is offline
4. Write more files with one missing drive
5. Unplug the other drive without telling nothing to BTRFS while system is offline
6. Replug the first unplugged drive without telling nothing to BTRFS while system is offline
7. Write more files with one missing drive
8. Replug the first unplugged drive without telling nothing to BTRFS while system is offline
9. Now depending on what drive i use for mount command i can see files written on step 4 or on step 7 and save them on a different drive (third drive)
10. Do a re-build (BTRFS balance) and put back that files that are now missing
11.- No data lost
So the raid must be done at file level, each file must not be saved on both drives (part of it on one drive, part on another), since doing so, when one disk is not plugged such file would not be readable, but it was.
I had not see any other kind of RAID0 like that, all the rest if one drive is missing you loose all.
The bad side of BTRFS raid0 does not improve write & read speeds if not asked to do it in different files, at least is what i had seen.
That different way of 'raid' is what make me ask how really works Windows Storage Spaces at low level... since i am planing to use three SSD on an old PCIe Gen 2 16x (motherboard port) with a non hardware raid PCIe Gen 3 x4 SATA III controller, motherboard SATA has only two ports and both only SATA II... and for big files (virtual machine disks in inmutable state with differencing file in non auto reset) each between 10 to 30 gigabytes.
Thanks for confirming that Windows Storage Spaces works at block level (stripes) ad not per file, now the next question (mostly for performance reasons) what would be the size of such stripes (128KiB, 1MiB, configurable or not, etc) and i must investigate about ReFS instead on NTFS also, any tip on both is wellcome; also if it would be OK to use such for ofimatic files, etc. and a last point, stripe size is used as save unit on each drive or is the block size that is divided onto all disks (for Raid0)... if 4 disks and stripe is 128KiB would it write 128KiB on each drive or 128KiB/4 on each drive. Different raid implementations do it differently... some hardware raid0 controllers requiere 2^N identical drives, others canuse 3, 5, 7 and any N>=2 number of drives.
By the way, i had seen that drives must be >=4GiB to be able to be used on Windows Storage Spaces, but on Windows screen it does not say such thing, or at least i did not find where... my 3.82GiB (4GB) USB mass storage did not show on Storage Spaces but when i create a virtual disk of 4GiB it is shown... so that was my guess, at least 4GiB.
Thanks in advance.
@@androidlogin3065 Correct terminology is: "chunk" is block written to one disk, also called "Interleave" - default is 256KB. Stripe is all chunks of multiple disks written or read together (at once). You can set these at creation time of Virtual Disk. Details are in this video: ua-cam.com/video/it-1WtNrNr4/v-deo.html
@@VedranKesegicSQL Thanks a lot, i am not an expert and want to know.
256KiB for chunk size looks like a lot if i undestand it correctly thay 256KiB chunk for four drives in raid0 equivalent would be 1MiB per read / write operation.
I had added a comment on that video you suggest (after seeing it fully), since after seeing it i am more confused than before.
YOu are in for a rude awakening looking for speed increases from adding a drive in parity with actual SATA drives....; performance plummets from 110-140 M/sec sequential writes down to 32 MB/sec.... Mirrored? 120-130 MB/sec! But with 3 drives w/ parity? 30 MB/sec...! Yayyyyyyyy!!!!!!!!!
I love your videos, but when you make a list of "these are all myths" they shouldn't be just QUESTIONS. They should be "these are false statements, and I'll explain why". A "question" can't be true nor false.... it's just a question.
please dont do raid with windows
Because? If you refer to bugs it had N years ago, that was fixed. I have seen many production system using Storage Spaces with great success. No problems whatsoever.
I would agree.. I have been using storage spaces in my data centers for years. No problems.
Besides storage spaces isn't raid. For windows software RAID you don't disk management..
m.ua-cam.com/video/EVgqvhmR0-E/v-deo.html