I generally wouldn't visit the Hotsos Training Day, mainly because I've been away from home and work for long enough, particularly when you add the travelling time at either end, but this time I was determined to attend because Tanel was presenting.
It was a busy room with a very high percentage of attendees staying for the additional day. I suppose once people were actually allowed out of the office in the current economic climate, they figured they might as well pack in as much learning as they could. Tanel described it as one of the peaks of his career and I can understand that. It's an honour to present at the Symposium, so a whole day must be a pretty special privliege, but he's more than earned it with the number of good presentations and blog posts he's come up with.
It didn't get off to a good start, though, as there was a problem with the focus of the projectors, but that was soon adjusted a little and people were happy as long as it was good enough to read the text.
Speaking of text, he kicked off spending quite a bit of time talking about using the right tools for the job - sqlplus, basically - and how we can make our own performance more efficient before we even start looking at performance issues. Make no mistake - despite having a love of pictures these days, I'm fundamentally a command line chap who frequently finds himself doing the things Tanel talked about, which consequently made me chuckle.
- When diagnosing Oracle problems reported by others, I ask them to stop using TOAD, their JDBC application or whatever it is they're using and login to sqlplus. Once the problem is recreated there, I know it's a real problem.
- On windows I use the CMD prompt version of sqlplus too and can't stand sqlplusw (but I should perhaps keep that quiet because lots of others seem to like it )
- On Linux he spent a lot of time on RLWRAP and command line completion and there were some really good ideas in there that I don't use enough so I'm going to revisit them.
He showed some neat and entertaining tricks with colours that I took some photos of but they wouldn't really do them justice. I'm expecting him to have a movie playing from sqlplus next year
While talking about his Snapper tool, he drew the comparisons between it and ASH data, which he was pretty positive about and mentioned how ASH data is like a DW fact table with multiple dimensions that allow flexible and relatively complex drill-down into specific dimensions to help diagnose performance issues. That's something I talk about quite a lot when I'm teaching people about ASH.
But he came at it more from the angle of his own Snapper tool, which is cool if you've never taken a look and between that and the section on latchprof and latchprofx, demonstrated pretty effectively that these tools allow you to look at most problems and to limit your investigations to those that matter. For example, for someone who is known for his work on diagnosing latch contention problems (amongst other things) he was very careful to point out that you shouldn't become obsessed by latch contention unless it's causing a problem! He didn't stop there, though, next up was the utility that I know a few people have been waiting for - the Mother Of All Tuning Scripts, or MOATS.
I wouldn't want to steal any of Tanel's thiunder by blogging about it in too much detail, I'm sure he'll do that himself, but MOAT is a bit like a top utility for Oracle. Although other such tools exist, e.g. OraTop, I've had a quick look and I think MOAT is probably more extensive and only requires some PL/SQL and SQL. I was pleased to see him acknowledge that it was something that he worked on with Adrian Billington who is the guy behind oracle-developer.net and someone who I've enjoyed a few beers with since I started working in London again.
Next up were SQL Performance Tools and thank goodness he re-emphasised that using the EXPLAIN PLAN FOR command is asking for trouble, particularly as there are far better tools these days to look at the plan from the child cursor instead. (My TOAD equivalent for this in the office is to berate people for using that bl**dy ambulance! LOL)
Unfortunately there were some growing Production problems at work that demanded my attention
so I had to bail out early to work in my room which meant that I missed
most of the SQL Tuning and Visualisation sections but I knew the Perfsheet content
already and you can count on me revisiting the whole day once I'm back
in the office. I particularly liked his course notes - not too detailed
to be unwieldy, but just enough detail to remind me of the day.
Although it was a really enjoyable day, I obviously knew a lot of the content already, partly because I have similar interests and partly from reading Tanel's blog. It made me wonder why people who go to Hotsos don't read blogs and stuff
the rest of the year, because I sensed a lot of the content was completely new to some others in the room. I suppose now they've found his blog, they might start.
After a few hours working in my room, things were sorted out and I could go to the bar across the street for a burger and a few beers with Marco. As soon as we walked in, Tanel shouted over and there were was an interesting little crew in there, with Tanel basking in the glow of a job well done. It was a shame about the incredibly loud karaoke or I might have stayed longer but it's safer to keep me away from karaoke - although I was quite looking forward to the Estonian version It was a good evening, though, and ended up with quite a few of us in the hotel bar. My main memories are of me defending 'my' pictures and Tanel and I praising Mr. Billington to the heavens.
The "Adrian Billington must be allowed out to a conference" campaign starts here! We won't take no for an answer.