• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Rebuilding storage server... am I crazy?

F

farscapesg1

2[H]4U
2FA
Joined
Aug 4, 2004
Messages
2,658
So, I built this system originally from used/spare parts and I've come to the conclusion that I wasting way too much electricity/heat for storage I'm just not using. That and I've got a couple drives that are dying. So, I'm looking to consolidate and trim things down.

Currently, I've got 3 pools configured (OpenIndiana + Napp-it);

1) 8x2TB drives running in a RAIDZ2 array. This array provides my fileserver needs (videos/music/photos/documents/etc.) via SMB, as well has having a 1TB thin LU shared via FC for VMWare datastore access (larger VMDKs like WHS2011 backup drive, System Center 2012 storage, etc.). All said, out of the @ 12 TB usable space I'm only using @ 5 TB.

2) 4x750 GB drives in a striped-mirror providing a 1 TB datastore for VMWare via FC. Only using @ 250 GB

3) 4x1TB drives in a striped-mirror providing 2 TB storage for my HTPC server (virtual machine running Win7 and SageTV) via iSCSI. Currently it is using @ 300GB, but it fluctuates and has used up to 1.5 TB during peak seasons (especially the Olympics when my wife wanted to record EVERYTHING!!!).

So, all told, I'm currently using less than 6TB of space.... when the server is housing @ 15TB usable space. Seems like overkill to me ;)

My thoughts are to replace the RAIDZ2 array and the 4x1TB drives.... with 6x3TB drives in stripe-mirror pairs for @ 9TB usable space. This new array should give me better IOPS and make it easier to expand if I need to (though, I've stayed around the 5TB mark now for file storage for the last 2 years) just by adding another pair.

This new pool would then serve PVR storage, file server, and secondary VMWare datastore duties. I could replace all my current disks with just these 6 (plenty of space), but I'm hesitant to put "all my eggs in one basket".. even though I am backing things up to a DroboPro and offsite to Crashplan (at least the important stuff). Ideally, I would rather replace those 4 750GB drives with a couple SSDs and be done... but that will have to wait for more money ;)

Am I crazy for thinking this setup should provide as good, if not better, performance with half the disks? Obviously it isn't the most "space efficient" option...
 
Correct me if I'm wrong but less drives will give you less IOPS.

I was in a similar boat as you in regards to my storage. I was using 17 2TB drives which sounded great until I started thinking about the power consumption over time. I went ahead and replaced them with 5 4TB drives.
 
I don't think it is ever cost effective to replace small drives with large.

I estimate $5/year to run a hard drive.

There are certainly reasons to replace small drives with large drives.
 
danswartz - yeah, I currently backup my important files on a nightly basis, as well as all my VMs (critical and ones that see a lot of change on a daily basis, ones with little change or unimportant changes weekly).

twistacatz - I definitely don't consider myself very familiar to storage. My understanding is that with a striped-mirror (similar to RAID10) I should get IOPS equal to 1 drive per mirror, so a 3-mirror setup would be approximately equal to 3 disks for IOPS. Conversely, a RAIDZ or RAIDZ2 (similar to RAID5 or RAID6) would have IOPS equal to a single disk... plus a little for additional drives. I think RAIDZ or RAIDZ2 would provide better sequential reads than striped-mirrors, but even the striped-mirrors should be more than enough to saturate a GB NIC.

There would also be other benefits, like drastically shorter rebuilt times if/when a drive fails and easier expansion and easier expansion if needed.

GeorgeHR - Yeah, I normally wouldn't replace the drives just for cost effective reasons. Right now I've got 2 disks that seem to occasionally drop out of the RAIDZ2 pool. Reseating and clearing the errors gets them going again for a couple months before it happens again. Since everything is backed up, I've been limping along until I could get new drives. Currently have 4 3TB drives sitting here now, about to order 2 more.

Honestly, I've had some issues with the RAIDZ2 pool anyways. For example, the pool has to SMB shares, Storage and Private. I use the Private share as a download target for Uttorrent/SABNZBD downloads. After a file is downloaded, sometimes when I try to cut/paste it from the Private share to the Storage share... something times out and the file gets lost. Windows returns an "invalid file handle".. and the file is just completely gone. Sort of irritating considering I could have just cut/pasted 6-8 GB of data before. Occasionally Utorrent will also seem to "lose connection" to the datastore causing downloads/seeds to just stop. The drive is still there, and I can open and browse it, so it seems to just be a momentary loss..
 
twistacatz said:
Correct me if I'm wrong but less drives will give you less IOPS.
Click to expand...
Yes, this is true. But ZFS is different. Less vdevs will give you less IOPS. So for ZFS it is better to have mirrors, than one raidz2 if you want IOPS.

ZFS reads will be fast as the number of individual disks, but writes will be the number of vdevs.
 
You must log in or register to reply here.
Back
Top