Jump to content
RESET Forums (homeservershow.com)

WHS vs. raid; 10+ disk systems - Theory


stalni
 Share

Recommended Posts

Discussion so far has been related to data protection. What about performance? What about setting up those 10 drives in 5 two drive striped arrays. This would increase read/write rates and allow you to protect your data via duplication. Obviously the more drives in the striped array the better the performance; three arrays with three drives each would be awesome as it would get you read/write rates in the low 300 MB/s range. That would be smokin'.

Link to comment
Share on other sites

@stalni: I have not had any experience with RAID 5E/5EE/6E, and they seem to make sense from the performance or load distribution standpoint, but I can't speak about their reliability without testing first. I agree with your 16 drive limit for RAID 6, but I normally use 1 spare only, as it already has significant protection. In practice, RAID 6 will more likely be limited by the number of drives your controller can support.

@ITTOG: RAID 0 arrays do improve performance directly, but they also have similarly amplified failure characteristics. A 3 drive RAID 0 is 3 times faster and 3 times more likely to fail than a bare drive. Nonetheless, that is certainly the most efficient way of enhancing disk performance, as long as the reliability issues are managed properly.

Link to comment
Share on other sites

Guest no-control

If you want a preformance RAID, then you're looking at the wrong setup. RAID0 2+ SSDs, VRaptors or 15kSAS.

My main rig uses a pair of 128gb SSD in a RAID0 and I come very close to SATA2 saturation. Failure rate of SSD isn't proven yet and they simply haven't been on market long enough.

Otherwise I'm with roddy...

Link to comment
Share on other sites

Roddy, given I use duplication I am not worried about failure in a raid setup any more than in a non-raid set up.

no-control, I am talking about all drives in raid, not just the system drives.

Link to comment
Share on other sites

ImTheTypeOfGuy(why not use only upper or lower case?) nested arrays containing raid 0, sound a little to dangerous for me in a backup scenario, and especially when considering the 1000/8=125 MB/s theoretical limit Gigabit nic's than a 300 MB/s disk read would not be interesting.

Link to comment
Share on other sites

stalni

The varying case helps with the separation of the words given there is no spaces.

As for the network limit, good point, but I would rather be limited by the network than the drives. So maybe two disks would suffice? Would just have to build it and then test to find out how many disks are needed to meet or slightly exceed the network limit.

Link to comment
Share on other sites

In general, even for large WHS systems, RAID 5/6/50/etc. seems counter-intuitive. For the enterprise, I've had problems in the past trying to increase storage size for file servers. Hardware RAID can be very tempermental about using like drives, and when you want to increase the ammount of storage, you either have to figure out how to move folders and mount-points around or you have to try increase the size of your logical volume with either Disk Management or some expensive and risky tool.

WHS takes a lot of the mess out of this. Since we're not managing high-bandwidth apps like a database or application server, I appreciate the simplicity WHS offers in terms of storage management for file shares. Where I feel WHS faulters, especially v1, is in the area of fault tolerance. I haven't had good experiences with folder duplication in WHS, so I opted to configure my server using RAID 1 mirrors. Each logical mirror is added as an individual drive in WHS, so as far as like drives, I only have to make sure I have at least 2 of the same drive. I don't incure the overhead associated with other RAID types, and I'm still free to use consumer-grade drives as opposed to RAID-enabled enterprise class drives. I also leave folder duplication turned off on all of my shares.

Overall, this setup has been good for my needs, though I do lament that I'm paying extra $$$ for fault tolerance on all my files even when I don't consider them important.

Link to comment
Share on other sites

mrossco,
Would be interested in knowning what bad experiences you have had with duplication. In terms all all around solutions (value, ease, reliability, scalability, performance), I still think it is the way to go. As all things, it is a comprimise. Won't give you the performance of RAID, but it less complicated and more scalable to use only when you need it. IMO.

Link to comment
Share on other sites

pcdoc, During both Beta and with the RTM release of WHS, I had significant issues with file transfers when Folder Duplication was turned on. File transfers that were taking < 1 hour on my old file server were running for the better part of a day on my WHS server. I was able to narrow the issue down to 2 factors: 1) Performance significantly improved when I turned Folder Duplication off, and 2) Performance seemed to suffer most when copying folders with lots of little files (i.e. source code) as opposed to folders with fewer but larger files (i.e. iso images).

When I finally got around to giving WHS another try, I opted for hardware RAID instead.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
 Share

×
×
  • Create New...