UNIX SIG 09/2009
Posted by Martin Bach on September 9, 2009
I can’t say I like junction 15 a lot … simply too much traffic when moving off the M25.
When arriving at the venue I was still quite early and had a chance to catch up with Phil Davis and Joel Goodman from Oracle.
It seems that the credit crunch affected the numbers quite badly as the turnout was really low. I think I counted roughly 20 delegates, what a shame. Maybe I have scared them off with my presentations?
This time the user group tried something completely new, and I’m not entirely sure if it’s a good thing. The afternoon sessions were recorded and broadcast via go-to meeting. In my opinion that reduces the value of a SIG meeting quite significantly as you no longer meet the presenters in person. I always found this kind of thing useful but took little pleasure in it. The oracleracsig.org site uses a similar scheme, but the sound quality usually leaves something to be desired.
This and some technical problems aside the day started with someone from Oracle giving the audience a demo of the new and flashy user interface for metalink, now rebranded “my oracle support”. It didn’t change the quality of the support though! The highlight was the announcement that Oracle will retire the apex based classical metalink soon which is great since most desktops, especially in high security environments will never use flash. This has become even more apparent as the presenter had to download the latest version of the flash plugin. This is something you could completely forget about in most corporate environments as all hardware is locked down. Well done Oracle! The comment that they are aware of the problem and working on a solution only prompted grins.
Next was my turn, and I seriously have to apologise for the delivery of the iSCSI presentation. When I first checked the time I only had 15 minutes left and haven’t even managed to get to the first demo, the creation of an iSCSI target! Bummer. I then had to step up the already fast pace to get towards the end and rushed the whole lot. I can only recommend that you download the latest version of the presentation to have a look – all the screen prints are on it and when looking at it at a quiet moment should convey the message. Next time I’ll rehearse more and cut the demo out, or try and get a 60 minute slot. Anyway.
Phil then presented the latest support update, few surprises here.
Over to David Kurtz, long time UNIX SIG chair. He presented about graphing AWR data through Excel. Previously I read that Tanel Poder had a similar tool, called PerfSheet.xls available to do similar things. Very useful, especially since you can point it to statspack tables for pre 10g databases. Although I have done business studies at the Uni I never felt compelled to use Excel so I wasn’t that fluent with the pivoting wizard – but definitely useful things!
After lunch, I was on to present about optimising Linux kernel parameters for Oracle. I have spent so much time preparing that presentation that I could impossibly fit it all into 45 minutes. Prior to the presentation I then hid a few more slides (which are still available online!) to make it shorter, the iscsi presentation disaster still vividly in my mind. I got a few nods from Nial Litchfield, which was encouraging. Joel then pointed out that the Oracle validated RPM would do a similar job, which is of course correct. But I wanted to show what’s happening behind the scenes… I don’t think there were too many kernel hackers in the audience, but despite that the presentation can still be used as reference (and I certainly will do so).
Row migration can aggravate contention on cache buffer chains latch – a war story and the second part of David’s presentation. He is seriously smart, and used his Excel graphs to find how row chaining (or migration – there was a lot of discussion :) caused a database system to hit a brick wall after the 80th user connected. Very good troubleshooting!
Mangled or Managed – Does Grid Control deliver? Nial’s presentation was all about Grid Control and his experience with the product. All in all a very good presentation and introduction to Grid Control’s architecture. He focused on the design, the good and bad bits as well as licensing (usually a minefield). Very round, but most of the topics were already known to me from my own experience with it during the OCM preparation.
That was it then – back home via M25 and M23, after a good 1:32 I reached Brighton. I had a bit of a mixed feeling this day, most likely caused by my messy iSCSI presentation and the low number of delegates. Let’s see how RAC SIG develops, I hope to see Piet and Jason there.