Brandon Fosdick wrote:
Erik Stian Tefre wrote:
You can avoid the problem by splitting the array up in partitions
smaller than
2TB each. (I know this does not answer your question, but it
simplifies things,
and it works for me(TM)... :-)
:) Thanks, but I thought of that already. This is going to be a big
database server and I don't want to have to deal with splitting the
database across two partitions.
If it's going to be a big database server, why aren't you using all of those
drive spindles to help break up the I/O load? :-)
Ask almost any Oracle or Sybase DB guru, and they'll ask for at least six disks
as three RAID-1 mirrors as a basic configuration, and would prefer 8 or 10 to
also hold the OS and the rollback logs on additional volumes. Blah, I tried to
point to some Oracle docs, but they're behind a registration-required section.
At the very least, you want to have your tablespace and your logs on seperate
spindles, ie, for a minimal config beyond just a single disk, you'd put the
boot volume and logs together, and have a second disk or RAID volume for your
main tablespace.
About the worst thing you could do to a database is put all of your disks into
one single RAID-5 volume. (Unless your database will be read-only.)
--
-Chuck
_______________________________________________
freebsd-stable@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
