CentOS 5.4 has been released…
I still use CentOS as the host OS for my home server so it’s nice to have the update, but I never use it for Oracle installations anymore. I only use Oracle Enterprise Linux (OEL) for that. Both RHEL and OEL have had 5.4 for a little while now, so there’s nothing different about the CentOS 5.4 release…
Howard Rogers noticed an issue with one of his utilities that is relevant to the majority of 10g/11g installation guides on the net and on Metalink. It looks to me like there has been a documentation change at some point. I’ve altered my 10g and 11g installation articles to comply with the current documentation recommendations.
It’s upgrade time again. WordPress 2.8.5 is here.
Here is a delayed final OOW post. The flight home was a nightmare. There was lots of turbulence for the first couple of hours. I got to the point where I just wanted us to crash and die because I was sick of waiting for it to happen. It all calmed down again, but every time there was a slight movement of the plane I expected it to start so it was not a pleasant flight in total…
I arrived home to find my central heating was broken, which is why I’ve not posted. It was too cold to use the computer. I just stayed in bed for a couple of days trying to stay warm and get over the jet lag…
The heating is sort-of fixed now… A guy is coming tomorrow to finish the job…
OpenWorld was noticibly quieter this year, which reduced the spectacle somewhat, but made life a lot simpler navigating round the place. I mentioned previously, I felt this year was my best OpenWorld. I think it took the first few visits to get my head into the right place to get the most out of it. The quality of the sessions in the unconference was amazing. I think most conferences would be happy if they were their headlining sessions. The unconference is probably the most important bit of OpenWorld for me now.
A big thanks to the OTN crowd for organizing the trip. Big thanks also to the ACE program for covering the costs. In the current climate, I’m not sure how long that will continue, but while it does I’m very grateful.
So, next stop is Australia for the AUSOUG conference series…
- Kevin Closson cancelled his session, which was a bummer, but life goes on.
- Oracle Optimizer Roundtable: Some members of the optimizer team answered the audience questions.
- Real-World Database Performance Roundtable: Members of this group answered questions from the audience on a variety of performance related subjects.
The roundtable sessions were pretty cool, but it feels a little like groundhog day sometimes. I think I could summarize much of the advice as follows:
- Design your database properly.
- Write good SQL. This is usually pretty easy if you have a good design.
- Use default settings and functionality, unless you hit an issue that forces you not to. Even then, don’t assume you can not use the defaults for other projects. Work on an exception basis, not a one-size-fits all approach.
- Newer versions of the database probably have features to eliminate or offset the problems seen in previous versions. If possible upgrade and make sure you understand the new features so you can apply them.
I don’t know about you, but this list has been true for as long as I’ve worked in Oracle. Unless you are working on systems requiring extreme performance or of extreme size, modern Oracle versions make life incredibly easy. It would be cool if people actually started taking the advice given by these guys every year…
Wednesday was a slash and burn day for me. I woke up and decided I couldn’t cope with any more sessions so I ditched the lot. Some of the RAC ones looked really cool, but I just couldn’t drag myself out of the seat to go to them. Instead I went to the OTN lounge to chill out and chat to people. Lots of cool people, talking about technical and non-technical issues.
I ended up tagging along with some of the guys to a couple of discussion type unconference sessions, but generally it was a downtime day.
The “big” keynote kinda passed me by. There was nothing there that held much interest for me, either because I already knew about it, or just didn’t care. The Arnold thing was a little surreal, but celebrities will do anything for cash, so I shouldn’t be surprised.
In the evening I went to Treasure Island with some of the guys. I ticked a couple of boxes (Seen Aerosmith and Roger Daltry), then came home.
I feel a little guilty about being so apathetic about the day when others would give there right arm to be here, but OpenWorld started nearly a week ago for me, because of the ACE Directors meeting, so I was feeling really burnt out.
Some of the data warehousing and Exadata presentations have talked about queuing requests to improve performance. They have suggested that using Resource Manager to throttle the number of active requests results in better performance/throughput compared to letting multiple requests all run simultaneously. The Terabyte Hour session yesterday showed an example of this and sure enough, when they limited the heavy requests to batches of 3, the overall throughput of requests improved.
This kinda links to something I keep banging on about in my PL/SQL presentations, which is decoupling. Don’t do it, queue it…
In any system you have a variety of business functions that have differing importance and differing required turnaround time. It’s good if you can identify this up front so you can consider decoupling some business functions. That way, functions that MUST happen instantly are fired on the spot, while those that can accommodate some lag time are queued for later processing. The acceptable lag for each of these business functions may vary.
Why do I care? Two reasons really:
- Why hog resources processing low priority tasks when they could be used for high priority tasks? You don’t run your backups and stats collection during peak hours. Why would you waste cycles on low priority business functions when the user experience is poor due to lack of resources.
- Decoupling allows you to take small transactions and batch them up, allowing you to take advantage of performance features available in both SQL and PL/SQL.
Queuing and decoupling are by no means new concepts, but they seem to have been lost in the mix. It’s interesting to see them being brought back onto the agenda, even with the sort of horsepower provided by Exadata.
Marcel Kratochvil recently wrote a paper on the relevance of FKs with respect to multimedia. This post is inspired by that paper, because it reminded me of a multimedia database project I worked on a few years ago, but it is not the central argument of his paper.
When retrieving large multimedia objects (Images, Audio, Video etc) from the database and sending them over the internet to a browser, the biggest impact on the user experience was the internet speed, especially since this was a few years ago. The database performance was for the most part irrelevant. As a result, the effect of bad execution plans was also largely irrelevant. So I might represent the loss of overall performance as a result of omitting FKs (or anything else that affects the optimizers decision making process) as follows.
In attempting to tune this system, you can bet that the majority of the effort was spent on the transit mechanism, rather than the DB performance.
- Am I suggesting FKs are irrelevant? No.
- Do I agree with Marcel’s paper? No.
- Do I understand some of the points he is making? Yes.
Your frame of reference has a big impact on the importance you place on things.
It’s late. I’m going to bed. 🙂
It was raining quite hard this morning. Fortunately I brought my “Fujitsu Siemens” umbrella from 2 years ago. Debra Lilley was quick to point out the company name change… 🙂
Today was another day heavily based around the Unconference in Moscone West. Four of the six presentations I went to today were at the unconference. Here’s what I saw today:
- What’s New in Eleven … Dot Two (that Oracle won’t be talking about) by Daniel Morgan: A quick romp through some of the stuff that is new or updated in the latest release that’s not sexy enough for the marketeers to focus on.
- Oracle Indexes: Q & A Discussion by Richard Foote: Richard was hoping we could all spend the hour discussing David Bowie, but the audience wanted to know about indexes. The session kinda turned into the Richard and Jonathan roadshow. Fortunately they agreed on most points. 🙂
- Under The Hood of Oracle Clusterware with live demo Part I & II by Alex Gorbachev: This was a 2 hour unconference session where Alex explained the architecture of Oracle Clusterware, then proceeded to demonstrate a whole bunch of stuff including scenarios causing node eviction.
- Current Trends in Real-World Database Performance by Andrew Holdsworth: I try to get to the sessions by the Real World Performance Group each year, just to make sure there’s not something new I’ve missed. It gives me a warm fuzzy feeling. 🙂
- The Terabyte Hour with the Real-World Performance Group by Andrew Holdsworth: This session involved a live ETL process loading 1 Terabyte of data in flat files into a Exadata V1 Database Machine running 11gR2. Once the data was in they gathered stats, did a bunch of transformations and issued a bunch of heavy SQL statements at it to show the performance. Very impressive to get all this one in 50 minutes. Nearly all the processing was CPU bound, so it would be interesting to see the same demo on the V2 machines. Even better to do the Pepsi Challenge and have then side by side doing it…
I’ve just noticed I’m about 90 minutes late for the blogger get together. I better get moving…
Despite all the geek fun, there is a downside to OOW. I for one will be visiting the doctor when I get back to the UK to ask for some Prozac. Why? I’m depressed due to multi-angle-envy. What?
- I’ve seen some excellent presenters whose technique is so slick it makes me sick. I’m not talking about fancy slides. I’m talking about how they engage the audience. Makes me feel like a rank amateur.
- The amount of technical knowledge some of these guys have is crazy. Now I know they are focused on very specific areas, but even so. It really brings you down to earth with a bump.
- I really, really, really want some of the kit being used by people out there. I want to run queries on multi-terabytes of data without indexes and still expect a result back within a couple of minutes. I need an Exadata V2 in my spare bedroom, along with a reinforced floor.
- There are so many products and features that look amazing and I know I won’t have the time, equipment or brain capacity to try them.
Sometimes OOW can really bum you out…