So why would anyone need a no-op storage engine? The answer is simple enough if you consider that writing the binlogs, which are the basis of replication, is done by a different part of MySQL (i. e. not the storage engine). So it is possible to set up an "intermediate slave" which applies the desired filtering operations on the incoming replication data from the master and writes new, much shorter binlogs itself. TheMySQL manual contains a brief description of this procedure. However it is far too short to provide ready-to-go instructions. I will try to do this here.
PS. I know it can't be a complete explanation in parts. Don't be a wise guy. Just enjoy. It's one of the smartest applications that I've seen.
No comments:
Post a Comment