April 2007 Archives

Woke up this morning at 7:30, surprisingly awake and aware, and got started on sessions for the day. It didn't last long.

The keynote I attended this morning was called 'Clash of the Database Egos'. They assembled the developers or heads of companies that developed the major and somewhat minor storage engines available for MySQL. They had Michael (Monty) Widenius, inventor of MyISAM, Mikael Ronström, inventor of MySQL Cluster, Heikki Tuuri, inventor of InnoDB, Jim Starkey, inventor of Falcon, Ari Valtanen, inventor of SolidDB, and Paul Whittington, inventor of Nitro. They had it set up like a type of battle royale, but most of them didn't really play along. Jim was probably the most colorful character up there, and I enjoyed listening to him speak about database and storage priorities -- I ended up going to a later session to listen to him talk about specific internals. He's very defensive and possessive of his technology. ;)

After the keynotes and a little work, I attended a session presented by Flickr on their 'Federation' technology. It's a homegrown system built up of 'database shards' -- master-slave relationships that allow certain users to exist on different machines, and replicating read info to other accessible machines. The whole thing is controlled by a few PHP scripts, and works very well for them. He did give away that most of the user-moves are a manual process, and a lot of manual observation is still needed. To me, it sounds overcomplicated in a lot of ways, but was a result of the limitations of standard MySQL Replication. I asked if the speaker had considered or evaluated other database engines before moving forward with the project, and he responded, "MySQL is the only database engine I am really familiar with, so no. I tried Oracle, but it was too complicated, so I gave up." The audience laughed in that "lol Oracle" way, and he moved on after that. I was left rather confused, as I probably wouldn't be caught dead saying "well, I didn't bother becoming aware of anything else, I just continued with blatant hacks." From what I gather, this was pre-Yahoo!, so it's not like there was a MySQL mandate. Even so, why not MySQL Cluster? He said that most of the IO happened out of memory anyway, it sounds like the perfect use of NDB. Note that I never advocate using Oracle.

I worked a bit after that, and had lunch in the outdoor patio. Ended up getting into a discussion on the limitations of MySQL Cluster and managed to make the guy think I was an idiot. Then he gets up and I see that he's an employee of MySQL on his badge.

Doh.

Not much else happened that afternoon as Brian and I got on a bus to head to the mothership, Apple, in Cupertino.

When we did return, I got to hear Peter Zaitsev, commonly known from mysqlperformanceblog.com, speak on MySQL performance hacks. It was a great session, he went very fast and made some big assumptions on the technical capability of those in the audience, but it was very easy to follow, and he gave some great insight on the limitations of MySQL and InnoDB and how to get around them. I will probably write more on it later, as I have to do a little research on my own.

We left the hotel to have some dinner, and decided to go to Birk's, a steakhouse right nearby. They had a lot of great ads, and looked pretty good. We hopped in a taxi, got there, and it was right between the two McAfee towers, and tasted like it's between two McAfee towers. Don't bother with Birk's. You'll be down a hundred, and no better for it. :)

Tonight was more Birds of a Feather Sessions, where I attended a get together on Falcon, as mentioned above. Fun note of the night was a group about FileMaker. From what I gather, they just drank.

I would too.

I took off from Seattle this morning somewhere around 6:40, surprisingly alert but mentally void. The flight was mostly uneventful, and I have to give Alaska credit, this is the first flight that left on time and arrived early. Coming into wonderful sunshine at San Jose International was a welcome respite from Seattle, especially after the winter we had.

Given the time of the flight, I arrived a little late for the opening keynotes for Tuesday, but I walked in about 15 minutes into Guy Kawasaki's talk on "The Art of Innovation". Guy used to be an Apple Evangelist, and currently works as a venture capitalist and runs garage.com. Now, I've been an Apple fan since the late 80's, as a little kid reading Mac magazines and studying the executive team and public players at Apple during the rise of the Macintosh. When I was 11 years old, I read a book of his, The Macintosh Way. Instead of being a book about the Macintosh specifically, as I was expecting at that age, it was really a book about the good and bad, the right and wrong of marketing. In a lot of ways, it taught me on the fine art of evangelism, targeted specifically at the Macintosh, but really managed to extend to most of the products and services I truly believed in. It made me a great retail salesperson in my late teens, and helps me extol the virtues of the products I use to friends and family.

I got off track.

Anyway, Guy spoke about innovation, and the art of creating great things and getting them out into the world in today's marketplace. He spoke about how people forget that you want to be "high and to the right", like our president. It's obvious, but seems to be lost on many entrepreneurs. Essentially, you want to be in the upper right of two axis: a product or service that provides great value to the end user, and doesn't have a lot (or any!) competition in the market. He drove home that you shouldn't let bozos bring you down, since they don't know anything, and to get your product out the door -- it doesn't have to be perfect right away, but get it perfect as soon as you can, as those eyes are going to be on your product.

This was probably the best thing of the day. As this is my first conference, I think I was expecting something far different than what I encountered. The sessions I went to, which sounded like concepts I really wanted to learn about, were things that I was already aware of or had experience in. I don't think I know a ton about MySQL, though I've developed against and administered it for about seven years now. These sessions erred on the side of the high-level talks, so I spent most of them working instead of being really involved in it.

Another thing I noticed is how heavy the PHP influence is at this conference. Most of the people here are beginner-type PHP developers, and most of the products and services are tuned to the PHP developer. Ruby on Rails was represented pretty heavily here, as ActiveRecord is targetted at MySQL. It really makes me want to think about doing a session or two on Web 2.0 and Catalyst integration, even if I'm not the biggest MySQL fan.

I am going to go pass out now.

I will be heading down, bright and early, to the MySQL Conference and Expo down in Santa Clara, CA tomorrow through Thursday. My employer is sending me down there, as we still rely on MySQL for all of our data warehousing, so I'm hoping to learn a bit more about enterprise-level scalability. I know replication fairly well, as well as tuning the database server for memory requirements and performance -- but I always seem to find out that I don't know as much as I think I do. If anything, I want to find out if InnoDB really sucks as much as I think it does, or if I'm missing something.

The one thing that bothers me is that no one from the perl community seems to care much about this conference. I know MySQL is a bit of a baby database, but there's little to no representation. Meanwhile, there are sessions for PHP and Ruby all over the friggin place. I wonder if I could step up to the plate? ;)

I've become more of a PostgreSQL fan as of late, and I have a feeling that $employer is going to be moving that direction as well, as three out of the five of us are gung ho for moving that direction. Clustering, scalability, and flexibility of deployment are all strengths of Postgres, and speed is simply not an issue anymore.

But, really, anything but Microsoft SQL Server at this point.

It's quite possible that I'm going to end up selling off my N800, only because what I do now negates most of the need for my N800. It also happens that my Nokia 6682 is starting to die, and I'm going to have to send it to Nokia for repair whether I keep it or not. So, what phone do I want next? I think I've narrowed it down to 3.

Nokia N76:
Pros: Thin, light
Cons: Standard keypad, easily mistaken for a gd RAZR

Nokia E61i:
Pros: Fairly thin, big display, qwerty keyboard, WLAN
Cons: Bigger than an average phone, can be slow

Nokia E90:
Pros: Huge display, huge keyboard, dual mode as open or closed, WLAN
Cons: Huge

Looks like the open source messenger "Gaim" is called Pidgin IM now.

I love corporate lawyers.

What a seriously ridiculous name. At least is isn't something non-sensical, like Avvanta. :P