Proactive page cleaning

When the number of good victim pages is less than an acceptable value, page cleaners search the entire buffer pool, writing out potential victim pages and informing the agents of the location of these pages.

When the oldest dirty page in the buffer pool exceeds the configured time for the page_age_trgt_mcr database configuration parameter, the database is said to be experiencing an LSN gap.

Under the default page cleaning method, a logger that detects an LSN gap triggers the page cleaners to write out all the pages that are contributing to the LSN gap; that is, the page cleaners write out those pages that are older than what is allowed by the value of page_age_trgt_mcr. Page cleaners alternate between idleness and bursts of activity writing large numbers of pages. This can result in saturation of the I/O subsystem, which then affects other agents that are reading or writing pages. Moreover, by the time that an LSN gap is detected, the page cleaners might not be able to clean fast enough, and DB2® might run out of log space.

The proactive page cleaning method modulates this behavior by distributing the same number of writes over a longer period of time. The page cleaners do this by cleaning not only the pages that are contributing to an LSN gap, but also pages that are likely to contribute to an impending LSN gap, based on the current level of activity.