Atom vs. Geode: Which Makes a Faster, Cheaper NAS?

[vnangia]

Not to hijack the thread*, but how was your experience with OpenFiler? I personally found it too difficult to use - and went back to my beloved FreeNAS, but I'm not sure if my difficulties were a PEBKAC-type issue. For instance, I never did get remote SFTP working, nor user logins.

On the other hand, the fit and finish of OpenFiler was astounding. FreeNAS still has a number of rough edges, not the least of which is the fact that it can't partition disks. I, for one, am scared to think what will happen should one of my disks fail and I buy a similar, but not identical, model.

*: Is it possible for a first poster to hijack a thread?

[thiggins]

I tried to summarize my Openfiler experience in the last page of the article.
So far, the combination of Ubuntu Server, Webmin and mdadm has produced
the best performance. Not the best user interface, but it works.

[blister]

Any thoughts on using the Via chips here?

[thiggins]

What would you suggest for an inexpensive Via board that would be comparable to the Atom or Geode?

Since the Atom, Geode, and Via are all x86, these NAS boxes should be able to run Windows Home Server. I am quite certain your Atom board could, but I am not certain about the Thecus N4100PRO (it would at least need a RAM upgrade), and I don't know whether performance would be adequate. How about installing WHS and running these tests again (and covering all the additional features of WHS)?

[thiggins]

I have reviewed an HP WHS system and also featured it in an early Fast NAS installment.

Might be interesting to see the performance on Atom, however.

Originally Posted by thiggins

What would you suggest for an inexpensive Via board that would be comparable to the Atom or Geode?

I don't think it was available when you did the comparison, but now there is the VIA VB8001-16 (using the new Nano CPU). Logic Supply has it, but I think I've seen it less from other suppliers.

If you do do something with this, could you please do a test with both 32-bit and 64-bit distros, if available?

Thanks
Steven Leary

Hey, just found this thread. I've been fighting with OpenFiler for about 2-3 days now and finally figured out a few critical things to getting SMB working, which are minor corrections to the posted article. I agree that it's shittily documented (because they actually make money by SELLING the administrator guide docs), but with enough persistence, it is manageable.

For one, you can set the windows workgroup on the ACCOUNTS page. I'm using LDAP, so that box is checked and contents of the LDAP configuration is filled out. The next section "Use Windows domain controller and authentication" is not checked, but if you put the workgroup name in "Domain / Workgroup:", it will respect it anyway.

Also, I had trouble getting windows to recognize my openfiler hosted SMB share at all. It seemed really, really sluggish and didn't want to respond at all until I forced it to Public Shared Access, and I poked at it at a command prompt with "net view \\openfiler". It eventually woke up and I could find the share. Once that showed up, though, WinXP Pro had no trouble accessing the lengthy \\host\volume.disk.share path.

Like I said, I'm still fighting with it, but I think it's pretty well done. I fought with Redmine/SVN installation for 2 weeks before claiming a victory, so if I can get OpenFiler up in less than 1 week, I'm happy. Although, using SWAT to configure SMB shares isn't all that bad, and seemed to work fine for me in the past... the main feature I will enjoy is FTP access to my SMB shares, which is awesome. That way I don't need to set up SMB on my various servers just to push out or pull in a file or two as needed.

JH

[vicx]

Firstly I love the articles and tests you have here. I have read most of them and have found them informative and rigorous (everything they should be).

About the Atom vs Geode article. I thought the article was rigorous as always but I got to the end of the tests and thought you had missed a chance to add something more to the evaluation of NAS devices with respect to the new more powerful x86 based device and the extra packaged software modules available for them.

I know that Via/Atom/Geode platform are less efficient/slower than dedicated controllers at the same price but from reading your tests I saw that they were 'good enough' at traditional NAS duties. What I was hoping to see was how much better they are at general computing abilities vs non-x86.

I can't recall seeing the general computing abilities of NAS units rigorously benchmarked yet. I can see that it hasn't been that relevant in the past for NAS units but I'm wondering if the environment is changing a bit. Seeing all the software available in the QNAP-QPKG repositories is an eye opener and other manufacturers and communities are also pursuing software packaging and delivery targeting these new platforms.

Just wondering if one or two tests on the general processing power available to run some of the more intensive software tasks would be acknowledging the area where the x86 based processors have some strengths despite their extra cost and power consumption. Maybe unpacking some RARS or indexing a few GB of MP3s with a media server would be a decent benchmark. This is just an idea I had after reading your article while waiting for my popcorn hour to finish unraring a few TV shows we don't get here yet.

All said I really love the site - its one of the best on the web.

[thiggins]

Thanks for the suggestion vicx and the kind words about SNB.

That's an interesting thought. Probably won't be something that I would add to the general test suite. But could be something for a special article.