Plans for dynamic configuration and other cloud properties?

Asked by Guerry Semones on 2009-11-18

I'm quite excited about the work going on with Drizzle. I'm curious about where the Drizzle cloud-friendliness plans are going in respect to a few points.

So many services that I work with in the cloud always expect statically configured files to point to peer nodes. Introducing new nodes or removing nodes almost invariably requires a restart. What I see that we need in a cloud database are things like dynamic discovery and membership management that self-organizes and self-heals without administrative oversight (except perhaps to introduce a new node or remove a node--and even then that should be an API call that can be automated). Taking that to the next level would be auto-sharding based on policy or some criteria to balance across the nodes. And, of course, dealing with lost nodes gracefully and automatically.

These things are important because, from my view, cloud is not a static environment, but one where services may be required to move around and run on different servers due to load changes, SLA, etc.

Are any of these things, or aspects of them in the Drizzle roadmap? I've seen mention that Drizzle is actually part of a larger effort called Pandora. However, I can't locate any information on Pandora. If I'm asking in the wrong forum, my apologies. Please point me to the right place.

Question information

English Edit question
Drizzle Edit question
No assignee Edit question
Last query:
Last reply:
Brian Aker (brianaker) said : #1


In our plans we want to pull out the configuration system and place it behind a modular interface. Once that is done we should be able to allow other plugins to exist (so... toss out the my.cnf approach if you want a server to configure itself from a different configuration source).


Can you help with this problem?

Provide an answer of your own, or ask Guerry Semones for more information if necessary.

To post a message you must log in.