Posted on 5 Comments

Mmm, what an interesting week

I have been very busy here in Malaysia this week.Β On thursday, I was asked to do a MySQL University session on MMM. The preparation was very stressful. There was no good wifi to be found until literally a few hours before the session (Big thank you to Gurdip at APIIT for providing a space and exceptional help!). On top of that, dimdim, the software used by MySQL for their sessions doesn’t seem to want to work on Linux (particularly the speaker part). I ended up using a laptop borrowed from APIIT with Vista and IE. Feels kind of counter-intuitive for a company in the FOSS business.

The session went very well and here is the resulting recording of the MMM talk on the mysqlforge page.

But that wasn’t the end of the MMM-promotion week:Β I happened to be allowed to present at the foss.my conference in Kuala Lumpur pretty last minute. At first I was going to do an updated version of the talk I gave at Froscon in August, but I was asked to do a tutorial session of 3 hours. I had never done anything like that, but I am always up for a challenge πŸ™‚

Again, preparation was a bit stressful. I didn’t know how many people to expect and it wasn’t clear if I would achieve getting running MMM clusters up in 3 hours. Well, I was underestimating my own capabilities apparently. Almost 100 people showed up, most of them without a laptop. I was surprised at that and explained them that it was probably not going to be so interesting for them. Again, I was wrong. While the laptop-owners prepared their laptops, I used my time to explain to everyone what MMM is, and how it works. Then we set up the laptops, solving all the problems we met on the beamer that we had a user connected to.

In the end we managed to set up 2 clusters within exactly 3 hours. Only 6 (almost 7) ‘servers’ participated in that end-result, for various reasons the rest was not possible. Still, it was a very good result and the attendees were visibly very happy.

If you hadn’t noticed yet, I’m a big fanboy for MMM and thinks this project needs/deserves a lot more visibility. It really solves a bunch of problems many MySQL admins struggle with, while providing a simple, cheap HA solution. This week has been very good for the promotion of MMM.

Along the way I also discovered that I really love doing this workshop and I hope to do many more like this. On that note: if you know of any conferences or meetings in the Asia Pacific area in the upcoming months, let me know and I’ll try to be there with either a presentation or a workshop!

Posted on 5 Comments
Posted on

Storage Miniconf Deadline Extended!

The linux.conf.au organisers have given all miniconfs an additional few weeks to spruik for more proposal submissions, huzzah!

So if you didn’t submit a proposal because you weren’t sure whether you’d be able to attend LCA2010, you now have until October 23 to convince your boss to send you and get your proposal in.

Posted on
Posted on

FrOSCon 2009: 1 down, 1 to go

So, I’m in my hotel room and I got some sleep. Now, it is 2:15 am and Icouldn’t sleep for some reason. PRobably has something to do with me being narcoleptic, as weird as it sounds πŸ™‚ Since I can’t sleep for a bit anyway, I figured writing a blog summarising the first day of FrOSCon would be a good idea.

Well, in one word: GREAT! In a few more: The first day was a big success. I started it out with a shift at the registration desk (I’m volunteering for FrOSCon as well). Since it was very early, most people that came in were actually exhibitors and speakers, and they all had to be at the VIP desk. At first, the door wouldn’t open. Later it turned out that the sensors thought we were to close to it. Claustrophobic doors are interesting.

Then, there was breakfast for the people with VIP access. Basically that was anyone who was not just a visitor. I helped out for a bit and then went back to my hotel room to catch up on some sleep and practice my talk a bit. The practice run was done in a staggering 57 minutes, which was exactly enough since I had 1 hour to finish. I went back to the conference and saw some talks. I saw the tail-end of Sheeri’s talk on ‘a better mysqltuner’ and then the start of Jan Kneschke’s talk on MySQL proxy. It was interesting, but my nerves made me exit that prematurely so I could be sure to be in time for my own talk.

When I walked into the conference room I was giving my talk in a half hour in advance, it was already empty because the previous talk was finished early. This gave me a decent amount of time to finish setting up.

I decided to use the wiimote that the CentOS guys were offering as a presentation remote. That turned out to be a very good decision, as it gives you something to hold in your hands so you hve soemthing to do with them. It also gives you force feedback and led information on how much time you have used for your talkso far. Pretty awesome!

The talk itself went pretty well, and people asked good questions afterwards. One guy came up to me an I thought he was asking me a question, only have hearing him sya he wrote this stuff. It took me a while to realise that it was Pascal Hoffman, the guy that wrote all of MMM 2. I hung out with him and my friend Nicolas from Rotterdam at the social event all night, speaking about many things related to MMM.

That is actually what these conferences are so great for: the human interaction aspect. You get to meet the humans behind an irc-name or blog, and talk to them directly, (almost) always a pleasant experience.

Before the social event I went to the PBXT talk in the OpenSQLCamp room, but i was too tired to focus on it with the attention it deserved. Kind of ashame, because i think PBXT is an interesting project.

Looking forward to tomorrow, in particular to the shootout in the OpenSQLCamp room between experts from various open source databases. I guess I should go back to sleep now, so I am full of energy tomorrow. Walter signing off!

PS. I’ll put the slides to my talk online as soon as possible.

Posted on
Posted on

MySQL conference ’09 Keynote: “State of the Dolphin” by Karen Padir

Here is a semi-liveblog from the opening keynote of the 2009 MySQL Conference & Expo on April 21st in Santa Clara, California. Karen Padir presents “State of the Dolphin”.

The room is big, with 2 screens on each side of the central stage. It seems to be completely filling up, but I got a good spot on the second row in the front. Directly around me I see Dups, Sheeri, Giuseppe, Kai, Mark Callaghan and many other faces I know, even they don’t necessarily know me πŸ™‚

Opening music, fancy animations. Quotes from satisfied customers, combined with tribal-style music and statistics.

Colin Charles kicks off. Good morning, welcome. learn a lot, provide feedback. Thanking sponsors in the first minute..

Introduces Karen Padir, she promised to bring more women to the conference next year.

Welcome, she introduces herself, tells where she is from and what she did in her education. Jokes: “Odds are good but the goods are odd” about sun’s crowd.
Went to Redhat for a year in between working for Sun, because of opensource interest. She came back to Sun because of Sun’s OSS commitment. Shows a slide with a bunch of different Open Source-ish projects at Sun.
She worked with Glassfish for a long time, encourages people to go by them at the expo hall.

Now, on Oracle. Talks about Oracles commitment to opensource. names two examples of sideline involvement of Oracle in Sun projects. Doesn’t really come across, but might be me.

Everyone uses MySQL everyday, whether on Facebook, Google or whatever.
MySQL had a great year! some stats fly by. 5.1 has gained momentum: 3+M downloads in 100 days. 54% of users have 5.1 in production.

Sheet: What I’ve heard. Waht she has learned in the past 2 months since she has been on the job.

MySQL 5.4 is released today. What happened to 5.2 and 5.3 http://www.mysql.com/5.4
– innodb scalability: up to 16 way x86
– subquery optimizations
– new query algorithms

Some stats fly by for InnoDb that look promising, but i wonder what will hapopen in real life.

Introduces Ken Jacobs. Ken starts talkng about where the improvements of 59% come from: opensource ecosystem.
5.4 will have more features in teh near future. Fast online index creation, compressed table storage. Will be intorduced in the next weeks.
Ken announces new product: embedded InnoDb. low level non-sql api.
karen gives him acquirer of teh year award, with a bottle of vodka. Whoever comes with 9 billion next year will win then πŸ™‚

Karen takes over again, introduces MySQL cluster 7.0. It disappears before I can look at the slide, apparently not that important πŸ™‚
Next: MySQL query analyser. She asks who uses it. very few hands. Pulls mark Matthews and Gary Whizin up on the stage with a joke. They work for MySQL Monitor engineering team.
They tell a story about what happens when you have a slow app. The oldway: dig in and have not so much fun. New way: enterprise monitor. Mark showcases the tool by analyzing an off-the -helf CRM, not very interesting.

Karen comes back, asks all former and current MySQL employees to stand up. She thanks them all, quite genuinely. Nice thought πŸ™‚

Community contributions. Says 5.4 is a good example of community contributions. Sounds strange since noone knew it was coming.

commits to monthly updates on all shipping community releases.

Invites Kaj Arno to the stage. Kaj says he wants to have documentation GPL’ed.

Next: MySQL drizzle project: she calls it shepherded by Sun/MySQL.

Next: MySQL awards.
Partner awards: Intel, infobright, lifeboat.
Applications of the year: Zappos.com, Alcatel-lucent and Symantec.
Community: Marc Delisle, Ronald Bradford and Shlomi Noach

Announces the rest of the conference.

Colin Charles announces Mark Callaghan (turns out he was sitting right in front of me πŸ™‚ ) from google. introduces himself.
Talks about adapting MySQL to their needs, easier than making apps adapt to MySQL. Talks about using MySQL at google, cannot name numbers unfortunately.
many primary shards, with many replica’s per shard. They are happy with MySQL.
Tells that he is amazed at how many slaves can be coupled with a master. They found out by trial-and-error.

Thanks, Monty, MySQL people, Heikki Tuuri and company. inspiration provided by Yasufumi and Percona.

History of MySQL at google: first MyISAM and another RDBMS. after that 4.0 and Inno. After mark arrived they upgraded to 5.0.

Talks abotu features of an enterprise DBMS, the problems they overcame, how they test new MySQL builds: stress test, sample workloads, killing off servers, check for compiler warnings and check for data drift (test same workload with old and new build and see if they are the same).

How to deploy: automatically searching aggregate error logs. Keep track of number of crashes on daily basis. Automated removal of machines from service.

How to monitor: user_stats and table_stats patches. Monitor top-down, they use more bash and awk then mysql enterprise monitor πŸ™‚

Talks abotu features they added (and sometimes removed) from/to replication. Crash safe-slaves, Semi-sync replication, global transaction ID’s
Performance features they added. InnoDb improvements, backport connection pool feature from MySQL 6. Stresses they are not the only ones doing it (Percona, Sun, Innodb)

Manageability features they added and other features they added: row-change logging, Flapulator (bash and awk scripts in cron jobs checking for crashes etc.), Online data drift checking.

Tells about the crises they have had and the open problems they have, the runs out of time.

Next it is on to Kickfire, who are a big sponsor. I end my liveblog here πŸ™‚

I will post pictures asapI’ve posted photos, although the camera was pretty crappy, so it might not be very good quality. Donations for a new camera are welcome πŸ˜‰

Zak’s uploaded the photos for me to his flickr account – view the set here.

Posted on