High Availability Panel Discussions at OpenHA Cluster Summit
Am Tweeting at the OpenSolaris HA Cluster Summit that precedes CommunityOne. Some really informative presentations and discussions. I particularlly liked the panel discussion with industry luminaries from Google, Sun, and Aster Data. Panelists emphasized that the general population is used to high availabilty for all kinds of data and services. They only stop and _think_ about availability when something isn't available. But not many people really understand what it takes to provide that availability. I really appreciated the comments about simplifying the user interface so that a user can digest what is happening and what to do when there is a catastrophic failure, but not overwhelm them with details. In the same vein, need to provide access to the dials and meters under the hood to the technical experts who need to diagnose and correct those catastrophic problems... give them the tools they need to help. (I'm thinking of the extenstive technical documentation I've seen that is meant to cover those technical details, but is probably not really consulted after major downtime of an HA system. This is really a software rather than a documentation problem.) I also found it interesting that the panelists thought that most customers don't really understand their own availability requirements.... Rather, they are budget-driven and focused on trying to address some vague problem with a fixed dollar amount.
