Oracle World Day 3


An exciting day today - but probably for the wrong reasons. A bomb threat was made on the Moscone centre, so it was pandemonium as 12,000 people tried to get out of the centre within a 5 min period. Highlight for me was watching an elderly gent try to shortcut his way up the down-escalator only to run out of puff half way up and slowly head back down to the bottom...

Before that occurred, a couple of very good presentations. Cary Millsap (ex-oracle support) described in detail the various quanta that comprise a raw trace file, with special attention paid to the errors that might occur in data aggregation with tkprof. Also there was a sneak preview of the service level trace collection thats coming with 10G - including automated trace collectionn and active session history accumulation - amazing details captured at 1 second intervals with minimal resource impact.

Acclaimed internals expert Steve Adams gave a talk on ASM (automated storage management). Very impressive stuff - basically obviates the need for an OS storage manager. Given a bunch of disks, ASM will automatically use direct kernelised asynchronous IO on the underlying oracle files, as well as guaranteeing dynamic load balancing across all available disks. Add a disk, and ASM automatically redistributes the files to use it. Also, because the redo logs are written synchronously, no dirty region logging is required which speeds up mirroring performance as well. Very cool stuff.