Travel to CloudCamp, No:SQL(East), ApacheCon, and more...

One of the things I get to do for Gear6 is travel to conferences. I get to talk about Memcached, and other stuff I find exciting, such as nosql, gearman, drizzle, and cloud computing. And meet with other technical and business people who are excited about this stuff.

Last week I was in Atlanta Georgia, to participate in No:SQL(East).

Because I arrived the evening before, I discovered that Cloudcamp Atlanta was going on. I sat down in the audience, and then started answering questions instead of asking them, and so got invited up to stand on the stage with the other volunteer experts there, and then was asked to facilitate a discussion section on "Databases in the Cloud". A lot of that discussion was about the newly announced AWS RDS, plus the expected questions about running MySQL and Postgres on AWS EBS, about using S3 as a NoSQL data store, and on running app servers, database servers, and Memcached servers on EC2 instances.

Later that evening, at the cocktail party, someone was explaining his need and idea for serving lots of static content media files off fleets of redundant cheap web servers. And I told him "Oh, that's MogileFS, that's already been implemented for you!".

That's one reason why these conferences and mixers are so useful, just to discover from each other all the things we don't know that we don't know.

For the NoSQLEast conference itself, it was a amazing array of presentations on new non-relational data stores. I was impressed by presentations on using the Dynamo concept to distribute app servers, not just data stores; on Pig and Cascading languages as higher level abstractions over Hadoop (I wonder if they can be retargeted to Gearman); on Neo4j for being so small and yet fitting its problem space so well; about Redis, which is like a more structured Memcached; and about tin, which provides a practical huge scale nosql data store, with NO syntax other than just the http protocol serving text files.

There was also a talk that was notable for redefining "NoSQL" as not meaning "Not SQL" but instead meaning "Not Only SQL". Which I like. SQL and the relational model still has legs, and will be with us for a long time to come, but it will no longer be the end all be all data storage abstraction that everything has to be shoehorned into.

And that's all just last week!

Currently I am at ApacheCon US 2009 in Oakland California.

Last night was the Oakland NoSQL meetup. While I was there, I discovered that there is about to be a new technical professional job: "Non-relational data architect" (which is about as silly as "non-mammal zooligist", but still...). I realized this when someone there gave me a NoSQL data problem, and asked my advice. I mapped out his problem, and determined that probably he wanted Neo4J plus a full text search system, and made a start at some of the node and vertex attributes he needed. He asked me where I could suggest he find a data architect who would not immediately try to normal-ize and table-ize the problem, and I really had no answer for him.

Today I learned about the new caching system in Apache 2.2. Currently implemented are "cache to filesystem" and "cache to shared memory segment". What is really needed is "cache to Memcached", especially in such a way that a fleet of related Apache servers could share cached items. That will be fun to see come together. Anything that lets app designers use more Memcached without having to do an app rewrite, the better.

I may take in a day at the Internet Identity Workshop over in Mountain View this week while I'm here.

And then next week is Open SQL Camp in Portland Oregon. Gear6 is a sponsor of Open SQL Camp.

And coming up are more CodeCamps, Cloudcamps, OSCon, the MySQL Conference, Velocity, the Web 2.0 Expo, Structure, LCA, ...

Whew!