Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

looking to build a box for holding 5 to 8 TB of storage using nexentastor OS. without using commercial solutions like netapp.

my current list is:

Intel Xeon

8GB of RAM

RAID controller with 256MB cache

what to choose for disks? can any sigle server host more than 5TB disks? Dell Poweredge? HP DL or ML Series?

share|improve this question
up vote 3 down vote accepted

Don't know about Dell but HP have the following;

  • DL360 - 8 x 2.5" in 1U
  • DL370 - 24 x 2.5", 14 x 3.5" in 4U
  • DL380 - 16 x 2.5 in 2U

Any of these can be combined with a number of the following disk shelves;

  • D2700 - 25 x 2.5" in 2U
  • MDS 600 - 70 x 3.5" in 5U

This last item allows you to have 576 x 3.5" + 16 x 2.5" for a total of 1.12PB in a single 42U rack.

Edit - but on a serious note just about any server will allow you to have 4 x 3.5" 2TB disks but if you want faster 2.5" disks they only come in 600GB sizes at the moment so you'd need 14 of them just for RAID 0 (28 for R10, 15 for R5 and 16 for R6).

share|improve this answer
He has this tagged as "zfs". That being the case, he'll need a RAID card that allows "jbod mode". Is that the case with the HP servers? Or perhaps he can just do a bunch of single-disk RAID0 sets? – EEAA Apr 20 '10 at 20:19
Some of them do, others not. – Chopper3 Apr 20 '10 at 21:06
2.5" has 1000gb maximum size at the moment - western digital ;) – TomTom Apr 21 '10 at 3:42
Yes 1TB if you want very slow (5200rpm, SATA), but 600GB if you want fast (10k rpm, SAS), and I was specifically talking about speed. – Chopper3 Apr 21 '10 at 6:10
No, ZFS doesn't need RAID at all, although it can interoperate well with it if you wish. Drives come in different speeds, from the slowest at 4200rpm to fastest at 15,00rpm - most desktop disks are 5400 or 7200, most server disks are 10,000 or 15,000 - generally the bigger the disk space the slower the disk. JBOD just means 'Just a Bunch Of Disks', meaning can each individual disk be presented to the OS as a load of disks instead of the controller presenting lots of disks as only one or a few logical RAID'ed arrays - ZFS usually expects disks to be exposed on their own. – Chopper3 Apr 21 '10 at 12:13

Look at SUperMicro - they can build you complete servers (barebone, a distributor of them will fill in the rest) that holds:

  • 24 discs in 2 height units (2.5"). THat is - hm - with 1tb discs, Raid 6... about 20tb.
  • 48 (!) discs in I think 5 units, 3.5" - calculate this out yourself for 2tb per disc.

In general Dell etc. suck for that - simply the wrong cases, not holding enough Discs. SuperMicro is the best I found (

share|improve this answer

Check the solaris or opensolaris HCL:

Do you want to go DIY? Check this:

For ZFS better don't invest too much money in RAID hardware and buy more ram instead.

share|improve this answer

"For ZFS better don't invest too much money in RAID hardware and buy more ram instead."

Second that! OpenSolaris and ZFS is a huge memory hog combination. Another consideration is also around ensuring that you have a more current AMD, if that is your processor of choice as it appears OpenSolaris is very "picky" with the older lines.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.