MySQL Enterprise Backup (MEB) supports backup and copyback of mysql database objects into/from different disks or tape drives. On backup, MEB reads the data from the disk, processes it in form of validation or compression(if requested by user) and writes to disk. These operations were performed in a serial fashion till 3.7.1, wherein the reading of data is completed, followed by processing, and then writing. The performance of this can be increased by pipelining the read, process and write operations to be run in parallel, and this is what exactly is done in the latest release MEB 3.8.0 so as to reduce the time taken for all kinds of backups and restore operations.
How does this pipeline parallelism work?
This pipelined parallel execution of read, process(if any) and write operations is achieved by invoking multiple threads for different operations and usage of buffers. Each operation is split into segments and threads …[Read more]