Colin Charles Agenda

We were at the Sun+Zend party last night, and it was a blast (thank you Jesse Silver!). If you’re a PostgreSQL or MySQL user/developer or just a general database geek, you should’ve been there. Why?

(watch the video if its stripped in your feed reader)

Monty Widenius (MySQL) and Josh Berkus (PostgreSQL), decided to start sumo wrestling! It ended with a 5-0 score, advantage MySQL.

An attendee Tim Moore twittered: “Postgres is totally losing the sumo match. I’m migrating all of my databases to MySQL tomorrow.”

Monty says, this is what we do to people that leave Sun! In fact, if you didn’t already know, Josh Berkus, my esteemed colleague in the Database Group at Sun Microsystems, is leaving his post as the PostgreSQL Team Lead. We met for the first time, face to face at foss.in last year, and all I can say is I’m truly saddened to see him leave. But thanks to the magic of the open source world, we’ll still be interacting, I’m sure. Good luck Josh! (and better sumo practising next time, mmmkay?)

Hi! So this is completely my notes taken from the conference, without my thoughts attached to it. I should definitely post a lot more about this, and how the community can “improve” in time. Just not today. Believe me, sitting in the talk, was highly painful, and I’m wondering where my aspirin stash might be. The slides will be available soon, and lets just consider this a learning experience. It reminded me of the time Eric Raymond came to the Fedora Project’s very first FUDCon in Boston 2005 (probably the only session without available video :P).

What MySQL can learn from PostgreSQL
Joshua Drake
(more…)

No, its not alphabet soup. Just some notes from the session at the Sun Tech Days. I’ve not looked at DTrace much (my only look into instrumentation, has been from SystemTap, which doesn’t deal with applications), but plan on doing so soon… I’ve managed to get OpenSolaris Developer Preview 2 installed in VirtualBox, so it can only start being more fun from here…

Want to learn more about DTrace and MySQL? Then come to the MySQL Conference & Expo 2008, in Santa Clara, California, because on Thursday, Ben Rockwood, from Joyent, will be presenting a session on DTrace and MySQL (read the abstract, its good). The talk covers the fact that you can get useful information currently, even without the embedded probes in current versions of MySQL. For more DTrace and MySQL tips, don’t hesitate to read Joyeur, Joyent’s weblog.

DtTace, Web 2.0, Java, AJAX, PHP and the rest (notes from the talk)
by Peter Karlsson, Solaris Technology Evangelist

DTrace now has providers for a large number of languages: JavaScript, Java, PHP, Python, Ruby. Perl is on the way.

MySQL 6.0 will have DTrace support; PostgreSQL already has this in Solaris currently. If building from source, there’s a flag that needs to be enabled. A lot of work was done thanks to a community member.

You need a Solaris kernel. Ported to OS X and FreeBSD. Supports “dynamic instrumentation”. D is the dynamic language, used to script instrumentation

Very common request? Find how much time is spent in a given function. The thread local variable (self->variable = expression;) - nowadays, you can be running two threads coming down in the same function call. DTrace - so this is great for multi-threaded debugging.

PHP doesn’t have DTrace integrated, so, get the Coolstack PHP.

DTrace probes have been added to Mozilla to help debug JavaScript applications. This is available in Firefox 3 (in beta now). There is also a generic DTrace framework, that isn’t just JavaScript only - the networking parts of Firefox, to look at how DNS lookups work, etc. all can be instrumented via DTrace.

Further reading? DTrace and PHP, demonstrated.

Technorati Tags: dtrace, php, javascript, mysql

Day 2 for me started with watching Simon Phipps talk about Sun’s FOSS Philosophy and Strategy. It rained in the morning, so the talk started a little late, and there were hopes of better attendance. Nonetheless, the talk was interesting, and the announcement that there was money in it for FOSS developers, was just fabulous. I took away a few points, which I ended up Twittering:

• There’s this idea of a global mesh nowadays, and its leading to a changing society. FOSS is all about it. And “Its Going Mainer Mainstream”!
• Investment in skills is important for any country. There should always be a preference to invest in the local workforce. Simon mentions that all this allows you to keep the sovereignty of your country.
• You cannot pirate free software. Want to avoid foreign interference, and all the worries of WIPO? Free software is the answer.
• Simon Phipps thinks “software patents are bananas”. I tend to agree.
• I also found out that the Sydney Opera House owns a trademark on all photos taken of the Opera House. That seemed retarded, and not long after, I found out this was similar with regards to the Petronas Twin Towers in KL. I can take a photo, but apparently, I can’t sell it on say, ShutterStock Photo or anything. Ridiculous.

I didn’t get to attend the next round of talks, mainly because I was giving my talk! The room was full, the questions were good, I was happy. I read a report, from Ditesh, so that’s a pretty good summary, I guess.

I really wanted to attend the Mozilla talk from Mitchell Baker (mainly because I’d have liked to have met her), however, I couldn’t resist going to the PostgreSQL 8.3 talk by Josh Berkus. It was an interesting talk, well rounded, with the occasional jab or two at MySQL. The attendance was about half full, and we had some unwelcome loud noises in the talk! I took away from it:

• Contributors are full participants. PostgreSQL is owned and run by the community. Write a patch, and its accepted? Be prepared to write documentation.
• CSV logging is now built-into PostgreSQL 8.3
• I was introduced to Heat Only Tuples (HOT). Benchmarking, then seems to be skewed towards greater performance gains in PostgreSQL
• MVCC: Overwriting model (InnoDB, Oracle) or the non-overwriting model (PostgreSQL, Firebird)
• The attention to standards is great. Extending SQL, to create the SKYLINE feature, to power approximate queries, however, this was rejected for the core of PostgreSQL, and is available in their foundry
• Release engineering in PostgreSQL is amazing. 6 weeks development, 2 weeks commit, and repeat.
• There are doubts of an embedded PostgreSQL - this is what SQLite is for. I like the focus of the core team here.
• You never want PostgreSQL running on handheld devices - heavy writes it has.
• “There is no one size fits all solution, for databases” — Josh Berkus. I tend to agree
• Why are few hosting companies providing PostgreSQL? Customers don’t ask for anything else? CPanel doesn’t run with PostgreSQL. pg_hba.conf (my.cnf equivalent) needs fixing, for controlling quotas (can be implemented via tablespaces), etc.

Next up, was the OpenMoko: What, why and how talk by Harald Welte. The talk was packed to the brim, and I didn’t learn much more than I’d have found out from their website. Its an interesting project, but with the upcoming Android, and the idea that I need a working phone now, I don’t know if OpenMoko is right. Besides, the battery life on that thing is horrid.

Colin Charles and Josh Berkus (photo, courtesy Josh Berkus)

Spent time talking to Josh Berkus, in the corridor, nearby the Sun booth about life, the universe, and everything. Then it was Lightning Talks, and dinner…

Technorati Tags: foss.in, foss.in 2007, sun, foss, strategy, philosophy, postgresql 8.3, simon phipps, josh berkus

I found the recent interview with Bruce Momjian (founder, lead architect, PostgreSQL) rather interesting. From it I took away:

• PostgreSQL has stringent quality assurance. This is because there isn’t the “luxury of putting out a bad release”. He mentions that in the world of open source, there is zero tolerance for things that don’t work; I however can find many examples contradicting this line of thinking. Release engineering is largely dependent on humans and they do make mistakes.
• “People are more confident with us that some of the commercial databases.” I believe this largely is how your company is run - tech-oriented or suit-oriented. Worse if you’re suit-oriented and largely public. Investors and upper management need to blame someone when things go wrong, and thats why support services are so great in the open source world. Accountability is key. The ability to fix customers problems is key. Going off on a tangent, Michael Meeks, distinguished engineer at Novell and OpenOffice.org hacker extraordinaire basically said:
  

  Ubuntu, claiming to ship and support OpenOffice.org, it’s a total joke - they have a part-time packager. At Mandriva, for example, the OpenOffice.org packager is a self-described ‘not a C++ programmer’. So how you can then go and say ‘we’ll support you’… Novell, at least, has people across the board working on the codebase, with a good understanding of lots of issues.

• Bruce mentions evolution - from stopping PostgreSQL crashing, to performance tuning, to enterprise features. He reckons that 8.2 is Enterprise Ready, and 8.3 and forward is going to include “revolutionary features that go beyond things you can’t do with other databases”.
• My favourite quote:
  

  “If you look in the next five years, PostgreSQL will be a poster child for databases period,” he said. “There is not really another database that’s enhancing at the speed of PostgreSQL, so what that would look like is hard to say.”

• Pretty bold statement, eh? No roadmaps, like most open source projects. I actually think that’s a plus point, because roadmaps suit suits, but are completely inaccurate most of the time.

Technorati Tags: postgresql, interview, bruce momjian, michael meeks, accountability, database