I am planning out a solution using master-slave replication, possibly with Pressflow 7, and wanted to do a reality check that the approach makes sense. I'd greatly appreciate your feedback, even though my use case is not particularly high performance.
My client uses a browser-based application to do data collection in remote locations. These 'satellites' have web access about once a week. Currently, they are dumping a SQL file, emailing it, and manually updating the master server. Ideally, the master and all slaves will be kept up to date with the complete DB as much as possible.
My plan is to:
- configure a MySQL master-slave architecture, with each satellite with a separate slave id.
- rebuild the data collection app with Drupal 7, and take advantage of Pressflow 7's support for mysql replication.
- When the satellite machine has web access, manually trigger the replication operation, or auto-detect connectivity to run the operation. Issue a notification to the user when the replication has completed.
Seems pretty straight-forward. Am I overlooking something?
Comments
replication delay
This sounds like a good thing to test in Dev Ops environment WAY before rolling it out...
HINT:
Track replication delay for every slave so that Master point in time is in sync with slave in a tolerable way.
Learning when delay is within tolerable limits is often a matter of technology and client support.
Jeremy Donson
Database and Systems Engineer
New York City