Re: What exactly is postgres doing during INSERT/UPDATE ?

Aidan Van Dyk <aidan@highrise.ca>

From: Aidan Van Dyk <aidan@highrise.ca>
To: Merlin Moncure <mmoncure@gmail.com>
Cc: Greg Stark <gsstark@mit.edu>, Mark Mielke <mark@mark.mielke.cc>, Scott Marlowe <scott.marlowe@gmail.com>, Luke Koops <luke.koops@entrust.com>, Joseph S <jks@selectacast.net>, "pgsql-performance@postgresql.org" <pgsql-performance@postgresql.org>
Date: 2009-08-31T14:48:09Z
Lists: pgsql-performance
* Merlin Moncure <mmoncure@gmail.com> [090831 10:38]:
 
> I agree, that's good analysis.  The main point I was making was that
> if you have say a 10 disk raid 5, you don't involve 10 disks, only
> two...a very common misconception.  I made another mistake that you
> didn't catch: you need to read *both* the data drive and the parity
> drive before writing, not just the parity drive.
> 
> I wonder if flash SSD are a better fit for raid 5 since the reads are
> much cheaper than writes and there is no rotational latency.  (also,
> $/gb is different, and so are the failure cases).

The other thing that scares me about raid-5 is the write-hole, and the
possible delayed inconsistency that brings...

Again, hopefully mitigated by a dependable controller w/ BBU...

-- 
Aidan Van Dyk                                             Create like a god,
aidan@highrise.ca                                       command like a king,
http://www.highrise.ca/                                   work like a slave.