Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

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

I'm starting a project to consolidate a bunch of data and make it available on our LAN. I'm thinking that we may have a need for anywhere from 100-200TB at any one time.

What factors should I be taking into consideration in deciding between SAN and NAS, technology choices, and vendors?

(Although I know shopping questions are off topic, I wouldn't say no to any suggested vendors and models as a starting point for my research. :) )

share|improve this question

closed as not constructive by EEAA, ewwhite, Tom O'Connor, Zoredache, warren Feb 27 '12 at 17:46

As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.If this question can be reworded to fit the rules in the help center, please edit the question.

Marc: I edited your question quite a bit, partly because it was originally a "shopping question" (see the FAQ) and likely to be closed, but also because I thought Nex7's answer was very good and I thought I could make the question fit that answer. – Ward Feb 25 '12 at 6:28

I concur with the NexentaStor suggestion, but since I work for Nexenta I suppose that's expected. Bear in mind no matter whom you go with, big storage (which 200 TB would qualify as, if on the low end) really requires some grasp of your use-case(s) and a discussion with a qualified storage engineer if you don't want to end up with something that is at best over-powered for your needs, and at worst far under-powered for your needs. There's a discussion to be had about what you intend to do with it (all uses), and where in the triangle you sit.

The triangle is the big 3 parts of storage, and the old saying goes (well, there's many variations of it, but here's one):

Capacity, Performance, Redundancy -- pick two.

No matter who you go with (and I always recommend shopping around), talk through your use-case with someone there, and make sure you get a SAN that's suited for your needs (and when in doubt, always err on the side of caution; a slightly over-built SAN is a good problem to have, as opposed to an under-built SAN and the terrors it can bring you).

I'd also suggest that unless you have a compelling reason for it, 1 system holding 200 TB may be a bad place to start. Perhaps go with 4 systems each holding 50 TB, or 2 systems holding 100 TB each, as safer (failure doesn't lead to 100% outage) and higher performance alternatives.

share|improve this answer
Thanks - that gives me some great starting points to call my regular IT vendors with next week. – Marc Feb 25 '12 at 17:03

The decision between SAN and NAS is whether you want the file serving to be done on the storage or the servers you attach to it. If you choose SAN, you will have users accessing files through normal file servers (whatever flavor you use), which access what they see as a hard drive through some sort of SCSI based block protocol. If you go with a NAS, you will get files served directly from the same device that manages the hard drives.

NAS products tend to generally be less sophisticated when it comes to hard drive management. Also, they tend to be not as good at serving files as a file server would be. They're either more expensive or less scalable. Since disk scaling and file services scaling are accomplished by upgrading the same box, you risk running out of headroom on one but not the other.

SAN products tend to focus on the disk. They'll have better performance and redundancy/reliability on the disk side, but file functionality (if it exists) will be less good than a NAS or a file server.

There are a couple of integrated boxes that started off life as a NAS or a SAN and now do both, with varying levels of success.

share|improve this answer

I would build it for myself from the following components:

It will be significantly cheaper than any other NAS solution and you may use any of your Linux distro + file sharing service on it.

I am using this configuration with 24 disks, rocks like s**t

Btw, if you have to think whether you should use SAN or NAS then use NAS. SAN has significantly higher price because there is a big difference between them.

SAN is good if you want to have a smooth disk throughput, NAS if you have plenty of data to share on the network. (ofcoz if you own the central bank, you may use a SAN as a back-end of your NAS)

share|improve this answer
You're Insane. I wouldn't trust 100-200TB to my own hand-rolled hardware and software pipeline. You fork out a $LOT of money to a vendor for their liability. That's how it works. You do not do this kind of thing yourself. – Tom O'Connor Feb 27 '12 at 17:38
And when it broke, you'd fix it with your components too? What about when it started acting funny in a way that didn't point to hardware failure, but wasn't obvious? Pro tip: storage vendors aren't making money for doing something you're better than them at. – Basil Feb 27 '12 at 17:41
I'm fairly sure you've greatly mis-understood the difference between a SAN and a NAS as well. It's more than just reversing the letters in the name... – Mark Henderson Feb 27 '12 at 19:34
Mark: I think you should read my answer again :) I suppose I know the difference between SAN and NAS. That is why I told a SAN can be a back-end of a NAS. NAS acts as a fileserver on a network, most of them provide NFS,SMB/CIFS,FTP and other sharing functionality without the requirement to have the client to involve an FC card or external SCSI. SAN provides a native SCSI support for you, most of them via FC HBAs in your server. The biggest advantage of SAN, it provides a reliable throughput, no matter of filesize or random reads. – Gabor Vincze Feb 28 '12 at 12:33
Basil: Yes. I am pretty sure that if something goes wrong, the vendor will not help you to recover your data, and as most of them use the same hardware configuration that you can build for yourself, you do not need to pay for a brand like "Isilon" or more. Ofcoz if you do not have the competency to repair a dead server, you should choose something with warranty for the whole system. – Gabor Vincze Feb 28 '12 at 12:38

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