Meeting Tuesday, November 15: Managing High Volume Writes

The BWPUG isn't dead, it's just been resting. After skipping a few
meetings because of conflicts with the fall conference season we're back
next week.

Administration note first: after testing out Meetup for a while, the
page at
http://www.meetup.com/Baltimore-Washington-PostgreSQL-Users-Group/ is
going to be the sole place to get announcements about future meetings.
We expect to shut down this mailing list after this message. If you'd
like to be notified about the BWPUG, please join that group and you can
get notifications there.

This month I'm going to present a talk on the issues around use
PostgreSQL for high volume write applications. PostgreSQL has worked
well for many OLTP style application, where a number of clients are
writing heavily to the database throughout the day. But if you combine
heavy write volume with archiving, eventually you'll start approaching a
data warehouse sized database, with terabytes of data. Designing and
tuning PostgreSQL for a combination of heavy incoming data and a large
archive size requires some specialized knowledge, from how the database
cleans up after old data to how partitioning can be used for this purpose.

Topic covered will include:

-Basic database tuning for high write volume
-Monitoring buffer cache throughput
-Issues with empty cache performance
-Adjusting VACUUM cleanup
-Partitioning for time series data
-Hardware for write-heavy database workloads
-Write scaling with Skytools

When: November 15, 6:30PM.
Where: 7070 Samuel Morse Dr, Columbia, MD, 21042.
Host: OmniTI