• btrfs - BTRFS is pretty awesome, with filesystem level snapshotting and compression, it promises to make some waves on the server and small devices. Unfortunately, its still marked as EXPERIMENTAL by its own developers, and there are known bugs. However, you can choose to play with it in Ubuntu 10.04, which should be helpful for people finding and submitting bugs so the developers can feel better about people using it. There is a desire to have it as the default filesystem for the next Ubuntu LTS release, which is pretty exciting.
  • Monitoring is too easy - Any time I see 10+ implementations of the same idea, I figure its probably something that is easy enough that people tend to write their own instead of searching for a solution. Monitoring and graphing seem to be in this category, with many solutions such as nagios, opennms, zenoss, munin, ganglia... the list goes on and on. We talked a lot about what to do in Ubuntu Server to make sure this is done well and makes sense, and basically ran out of time. The best part of the session though, was that we decided to focus on solving the data collection problem first, so each server takes responsibility for itself, and then allow centralized aggregation on another level.
  • Server Community - There is some desire to have people test Ubuntu Server before a release, especially for the LTS releases. A beta program was proposed, but there is some doubt (my own included) that this will actually get people to test before the .0 release. Basically I have to think that as a server admin, people aren't interested in even trying something in an unstable state. They'll take the .0 and build a new server rev, but they're not going to go around upgrading stable servers. This needs more thought and discussion definitely.

Sitting in the first session for Wednesday now listening to a session about the next 6 months of Ubuntu Enterprise Cloud and Eucalyptus development. Very exciting stuff!