[geeks] ZFS sanity check

[Phil Stracchino]

Hey folks,
I have a new server to bring up, and no prior experience with ZFS.  I'd 
like to pick your brains a little.

The new box has twelve 300GB SATA disks.  I'm going to need a non-ZFS 
boot volume for Solaris 10.  So for starters, I'm going to need a slice 
of each of the first two disks for mirrored boot.  How much disk space, 
realistically, am I going to need for Solaris 10?  Does this have to be 
UFS, or are there other options?  If so, are any of them better?

Then I'm going to need some space for swap, of course, and then allocate 
the rest of the space for ZFS.

That's where the questions come in.  Can I throw all the remaining space 
on the first few disks, plus the whole of the untouched disks, into one 
big zpool and RAIDZ2 across the lot?  Or do I need to group physical 
volumes of the same sizes and add the groups to the zpool?  I know my 
boot volume can't be ZFS yet.  Can my swap be on the zpool, and is it a 
good idea to do so?  Am I better off to put the swap on separate 
spindles from the boot disks to spread accesses, or put it on the same 
spindles to minimize the number of disks ZFS is only partially managing? 
  What else should I know?  What other factors should I be aware of that 
can affect my filesystem planning?  Is there a better way of approaching 
the whole problem?



-- 
   Phil Stracchino, CDK#2     DoD#299792458     ICBM: 43.5607, -71.355
   alaric at caerllewys.net   alaric at metrocast.net   phil at co.ordinate.org
          Renaissance Man, Unix ronin, Perl hacker, Free Stater
                  It's not the years, it's the mileage.