Oracle OpenWorld 2010: Tuesday

Tuesday was a fantastic day at OpenWorld. The day got off to a great start with a bravura performance by Richard Foote who was hosting an Index Q&A unconference session. Richard’s knowledge is encyclopedic and he was easily fielding (and explaining extremely clearly) a range of questions around indexing, Top tip was in regard multi column indexing, put the first column with least number of distinct values first.

I think it is a bit of a shame that more people did not get to see this, I don’t think it’s helping having the unconference quite a distance away from Moscone. Tim Hall is thinking that it might be good for our health. I think I also preferred the otn lounge when it was back in Moscone West.

Followed Richard’s session with a cracking talk by Maria Colgan who was explaing the various steps seen in explain plans, she explained the various mechanisms very clearly, this was a jam packed session and there was a tonne of people standing as well. Found out a full index scan does not scan the entire index, while a fast full index scan does.

I then followed this up with Carry Millsap Thinking Clearly About Performance. Carry started by getting DBAs to shout out, and then getting the developers in the audience to shout. Showed a venn diagram with one circle labelled with DBA and a non-overlapping one labelled with Developer. He was emphasising that when it comes to performance, it really helps if the DBAs know a bit about software development and likewise if the developers know a little about how the systems actually work. Carry was then going through what he was saying were the 21 Axioms of understanding performance. Final slide showed earlier DBA/Developer venn diagram with the circles overlapping this time.

The good sessions just kept coming, and next was a really good presentation on ASM Cluster File System. Quite a lot of talk about Oracle Cluster File System – Cloud Edition which I think is a bit unfortunately named, it has nothing to do with OCFS2 as it’s all built on ASM and ACFS. Bit of talk about the new features and some fascinating little slips in the Q&A. Deduplication coming to ACFS. Some interesting stats mentioned, > 65% of RAC deployments are on ASM, but only around 20% of single instance deployments – was surprised it would be so high, though I suppose some shops might just mandate ASM throughout to have a standard filesystem. I then met the best system integrator in Bulgaria: Svetoslav Gyurov.

Next I had a fantastic chat at the demo booths with both the ASM team and with Stephan Haisley regarding online upgrading/patching. I really think the demo booths are a fantastic part of OpenWorld and you can really get the inside story of really any feature or Oracle product that you want. Lot of chat with the ASM guys regarding extended distance RAC, and what might or might not happen if both nodes fail at different times – bring up the last node to fail first!

Last session of the day was a so called “RAC Internals” presentation, which was more of a generic overview than anything really – and they failed to manage a Q&A session.

Day finished off with a fantastic meal at a place called Midi – really delicious food, though I’ve now woken up at 03:00am with a thumping hangover.


3 thoughts on “Oracle OpenWorld 2010: Tuesday

  1. Jason,

    “Found out a full index scan does not scan the entire index, while a fast full index scan does.”

    That’s a somewhat ambiguous statement. A full scan does scan the whole of the entire leaf level, in order, nominally one block at a time. A fast full scan scans the whole segment up to the high water mark (with the usual ASSM exclusions).

    Was the intent of the comment to highlight the fact that the full scan reads virtually no branch blocks – or is there something more subtle going on ?

    • Hi Jonathan,

      The emphasis was on saying that the full in “full scan” does not mean the entire index, whereas the “full” in “fast full…” does. It was all on what is the definition of “full”

      In fact this is the note I made at the time (typo and all):

      “full index scan does not scan entire index only enough branch bloks to find the leaf”.

      No, I think you have hit the nail on the head with virtually no branch blocks. I don’t think the fact the full scan still does the entire leaf was got across that well in the presentation.

      Thanks for commenting!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s