LVM or Openfiler or Nexenta or ???

[mc_bsd]

First off, thanks for all the great How-To's and post's I've found so far...

But, here's my dilemma:

They want me to build out a huge single logical volume over multiple drives connected over gig ethernet (instead of a standard raid chassis).

I was thinking maybe LVM over iSCSI or similar (they even suggested just hooking up a bunch of commodity NAS drives).

But they also want it scripted to find the new drives as they're connected (or to at least be a simple add via webmin or similar to grow the volume when they add new drives).

What's the best way to do this (besides buying a real SAN/Raid Chassis)?

Openfiler and Nexenta both look like they might handle it, but would generally require more setup than they seem to be comfortable with. Can they be happily scripted? Can LVM handle it? Is there some sort of auto-discovery thing that could work - maybe if we kept storage on its own subnet?

Thanks for your advice!

[falko]

You can do this with LVM. This tutorial might help: http://www.howtoforge.com/linux_lvm

[spyderdyne]

We have built 2 SAN servers for our campus and have settled on using Nexenta. Our previous iSCSI server is running Openfiler, and although it is very dependable it has a very limited set of file system features. Nexenta and the option for setting up thin provisioned ZFS and Samba shares is an ideal solution, but as you mentioned the iSCSI setup is difficult. Hopefully the next release will provide the features promised to implement HA as we have built 2 arrays of 3.16TB each and want to have an active/active redundant system.

Bear in mind that the complexity of these systems is what warrants the high cost of prebuilt SANs. We operate under a very limited budget, and therefore have to build all of our systems from scratch. It is important to remember that SAN system administrators are their own profession, otherwise it can be alot to absorb and can become very frustrating when setting these up the first few times.