[BBLISA] Looking for possible options to NetApp storage

[Daniel Feenberg]

On Tue, 1 Feb 2011, Edward Ned Harvey wrote:

>> From: bblisa-bounces at bblisa.org [mailto:bblisa-bounces at bblisa.org] On
>> Behalf Of Scott Ehrlich
>>
>> Now, say you have available shelving or drive bays.    NetApp
>> shelves/storage additions cost a lot of money.   Say you want to add
>> 40 - 200+ TB.   Say you have a need to add more (multi terabyte) disk
>> storage, be it from NetApp oir someone else.
>
> A couple of years ago I retired our netapp in favor of a sun server with
> solaris & ZFS.  We use it for a NFS/CIFS server.  For this purpose, solaris
> & ZFS are much better functionality than the netapp...  Although I'm sure
> there must be something netapp does better, I don't know what it is.
>
> For a newcomer to ZFS, I would recommend buying sun hardware and probably
> solaris.  Just for the sake of staying in a mainstream supportable
> configuration.  But an awful lot of people who are seriously into ZFS later
> switch to nexenta and silicon mechanics.

I'd really be interested in anyone reporting on their experience with 
Nexenta or Silicon Mechanics - we run ZFS under FreeBSD on cheap consumer 
level hardware and it works well for about a tenth the cost of the Netapp, 
but throughput isn't as good as the Netapp. Will Nexenta do better? On 
FreeBSD the ZIL isn't supported.

One of the nicest things about using ZFS under FreeBSD is that it is a 
real OS, and we can do housekeeping with any of the many available scripts 
and programs. We looked into the Sun 7220 and the sales guy said it did 
not support that, although I couldn't tell if he meant we couldn't do it, 
or that they wouldn't help us.

One thing to think about is just how reliable your storage needs to be. I 
know it is heretical to say, but the usual model for storage - an airline 
reservation system is a case where the amount of storage is small, but it 
is very costly to be offline. So that suggests using expensive, gold 
plated storage. But a lot of stuff stored around here is in very large 
datasets, for jobs that run for weeks on end. Those users would rather 
have enough storage, and have it offline a couple of times a year, than
have to deal with constant shortage. (Data loss is another, separate, 
issue).

See also

   http://www.nber.org/sys-admin/linux-nas-raid.html

Daniel Feenberg