Log Buffer #181: a Carnival of the Vanities for DBAs

Thanks to David Edwards for giving me the opportunity to host edit a Log Buffer.

I'll confess to a bias first off, as my experience is in Oracle (and as a developer), so my apologies if I misread the significance of activity in other areas. Comments are open, so feel free to correct and add.

I'm going to start with a few items on backup and recovery.

Bill Graziano at Sqlteamwebblogs learns and shares lessons from a SAN failure. Always useful to hear about real-life recoveries where things aren't as neat as a simulation or rehearsal. Oli Sennhauser on Shinguz asks whether you are trusting your backup and recommends practicing restores.
Buck Woody reminds us about the artifacts outside the database which we rely on , when he tells us not forget to backup your Master Keys.
Ronald Bradford shows how he ensures he doesn't lose information from his MySQL error log, while from his SQL Server blog Jonathan Kehayias is automating setting up his logs .


Looking to the future now and mySQL and noSQL were in the news as Twitter looks to a new architecture.
At HighScalability, Ted Hoff looks toward the end of the dominance of mySQL and memcache while Mark Callahan at mysqlha describes the pluses and minuses of mySQL, SQL and noSQL.

As solid-state storage pours on the speed, new bottlenecks turn up for Vadim Tkachenko from the MySQLPerformance blog, who points the finger at some InnoDB mutex's. From the same blog, the key-cache hit is autopsied by Baron Schwartz with a followup from Pythian's Sheeri .Similar discussions have absorbed many Oracle DBAs for several years.

On a broader, and longer term note, OptimalDBA Daniel Fink looks at whether software is (still) a good career path


With more immediately useful items, Thinkdiff.net's Mahmud demonstrates how incrementing variables within a mySQL select statement can be used to replace complex ranking function and Ronald Bradford is teaching Oracle DBAs about MySQL and shows us that READ COMMITTED doesn't mean the same in both worlds.

In the world of Postgres, Hubert Lubaczewski at depesz proposes using Text over VARCHAR(n) but anticipates less than whole-hearted acceptance. He gets support from David Fetter with VARCHAR(n) considered harmful but not from Leo Hsu and Regina Obe who write in its defense.

At LessThanDot, Ted Krueger is discussing mirroring SQL Server databases across the continent.

Still on SQL Server (but applicable to Oracle too) Brent Ozar thinks we should understand more about the Optimizer and stop hitting it with a big stick. Tuning is also on the cards for Charles Hooper as he details the intricacies of DBMS_XPLAN options.

Guy Harrison has been looking at memory management on VMWare, which will interest those turning towards virtualisation.

Finally, remember sharing is good, so I'll finish with Thomas Roach at OracleRant doing an un-rant like sharing of Grid Control scripts.