- High Availability and Scalability Patches from Google - describes changes made to improve SMP performance and make replication easier to manage. The changes include the SMP patches, global transaction IDs, hiearchical replication and binlog event checksums.
- MySQL Row Change Event Extraction and Publish - describes changes made to generate a row-change log on a replication slave that records DDL and all row changes from the replication workload. The log is complete, correct and easy to parse.
- MySQL Performance on EC2 - describes MySQL performance when deployed in a cloud, or using features typically used in a cloud (virtualization, network attached storage). It will describe the performance bottlenecks and improvements from the Google patch.
I have a talk at the Percona Performance Conference on Thursday at 10:50am. The title is TBD, but I have proposed the following: Life of a dirty page: the disk IO architecture of InnoDB and what can be done to improve it.