RE: Bigfile tablespaces and ASSM.....

I'm not saying that bigfile w/ ASM makes more sense than bigfile on NAS or CFS. Just that it makes sense, if you're using ASM, to also use the bigfile option, and not have to deal with adding datafiles all the time. (The database I'm moving is around 700GB total, lots of tablespaces, some as large as 100TGB, all w/ 2GB raw volumes, served up from an EMC SAN via VxVM.) I like the idea of having a "pool" of storage (the ASM diskgroup, in this case), that each tablespace can automagically allocate from as needed. The same could be achieved via NAS or CFS, and you could simply grow the filesystem, rather than the ASM diskgroup.

I'm not sure what the argument, if any, there is for ASM over NAS or CFS, or vice versa. Perhaps that it's free from Oracle, no extra licensing? But I suspect, whichever way you go, you're likely to trade one set of problems/bugs/headaches for another.

I don't see any strong compelling reason for or against ASM at this point.

-Mark

--
Mark J. Bobak
Senior Oracle Architect
ProQuest Information & Learning

For a successful technology, reality must take precedence over public
relations, for Nature cannot be fooled.  --Richard P. Feynman, 1918-1988


-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Kevin Closson
Sent: Tuesday, October 03, 2006 3:22 PM
To: oracle-l_at_freelists.org
Subject: RE: Bigfile tablespaces and ASSM.....

 


>	I'm setting up my test environement w/ 10gR2, three node Opteron

RAC running 64-bit RHEL4, w/ ASM.  With ASM, it makes a lot of sense to
have one datafile per tablespace, that autoextends, and use bigfiles.
That way, you never have to add a datafile.  You just monitor the free
space in the disk group, and add space at the diskgroup level, when
necessary.  But, bigfiles require ASSM....though I'm not sure why...it
doesn't strike me that those two features would be related....but
apparently, they are.

...please explain why BIGFILE tablespace makes more sense with ASM than,
say NAS of CFS? I can create a filesystem on a volume that I can grow
without interruption to 16TB today, much much larger in an upcoming
release.
With
Bigfile in NAS/CFS, you don't have to add anything at the Oracle level
(as is the case with ASM). You simply get another LUN from the storage
group, slap in into the volume, and grow the filesystem...the BIGFILE
just keeps on extending. Oh, that that, would not, btw, force a
redistribution of all the data in the old portion of the BIGFILE--which
ASM forces on you as you add disks to the diskgroup. ODDR should (must
really) be a per file option. Makes no sense redistributing old stale
data just because you add space.

As always, I'm looking for a problem that ASM solves that is actually a
problem on NAS/CFS, but this doesn't look like it. Mark ?
--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l