Re: AUTOEXTEND

  • From: Dennis Williams <oracledba.williams@xxxxxxxxx>
  • To: DGoulet@xxxxxxxx
  • Date: Tue, 4 Oct 2005 16:40:02 -0500

I have to agree with Dick.

I think DBAs tend to assume they have full control for storage space
and then end up taking the blame when they can't control it. In
reality, when the users need more storage space, you simply allocate
it. I ended up viewing it as a shared responsibility between the
users, the DBAs, and the system administrators. If you switch to
autoallocate, you need to get buy-in from all these parties.
    In theory you can get a runaway process that gobbles up all your
disk space, and I have never had that happen, even in development.
Again, if you make the responsibilities clear, then the fault lies
with the defective program, not you. And you just clean up the
problem, rebuild that tablespace and recapture the disk space.
    Originally I used to spend one day a week carefully planning data
file growth. But then I ended up with too many databases and too few
days. I could track the total storage by device much quicker than I
could track each tablespace. I suspect there are a few DBAs on this
list which are in that pinch.
    Think of Oracle storage settings as failure points. If you set the
storage limit on each data file, then you have that many failure
points. If you only set the limit with autoextend, then in theory you
have only a single failure point. I received fewer storage failures
after switching to autoextend.

Brandon - I've heard of the big time delay to allocate another extent,
but I have yet to observe it, or have a user complain.

Dennis Williams
--
//www.freelists.org/webpage/oracle-l

Other related posts: