Re: Multiple WALs for PITR ?

On Thu, Oct 07, 2004 at 10:32:04AM -0400, Goulet, Dick scratched on the wall:
> Some disk array manufacturers had a problem with their writes in that
> although the write the data was being stored in volatile memory instead
> of on disk. The disk buffers were volatile in that if power suddenly
> died all writes in memory that had not made it to disk were lost.
> Lately most array manufacturers (at least those I have experience with
> namely EMC & NetApp) have internal battery backup systems in place so
> that in the event of a power failure the array will accept no new data,
> but has sufficient battery power available to flush memory to a disk
> drive cleanly before crashing.

All of the RAID systems I've dealt with work a little differently. In
most cases the internal array battery is only there to keep the
contents of the buffer alive-- there is not enough power to actually
keep a disk spinning and write the data out. You'd need a real UPS
for that. Typically the internal batteries have enough power to
keep the volatile memory contents active for two weeks or more.
Once power is re-applied to the array, the buffers are
automatically flushed by the RAID controller.

They are also, frankly, a pain in the butt. We've had most of our
batteries fail within 18 months or so, and replacements are often
extremely expensive. For most of our applications we've just
disabled the RAID controller write-caches (our application writes
data at a very very slow rate; the main reason we have RAIDs is
protection, size, and *read* speed).

-j

--
Jay A. Kreibich | Comm. Technologies, R&D
jak(at)uiuc(dot)edu | Campus IT & Edu. Svcs.
<http://www.uiuc.edu/~jak> | University of Illinois at U/C