We are running into performance issues with Azure Sql Data Sync that the tracking tables are massive, and can be larger than the source tables (2-10GB for tracking tables). This makes the datasync very slow and db intensive. This is especially painful in logging tables where we can generate gigs of data each month.
Reading https://jtabadero.wordpress.com/2012/08/23/things-you-need-to-know-about-sql-data-sync-service/ item 17
The Data Sync Service periodically does metadata clean-up. This removes entries from the tracking table that are more than the retention period.
For example, there is no point keeping the metadata for deleted rows when these changes has long been propagated to the member database.
Currently, this retention period is set to 45days. That means delete metadata for rows deleted more than 45 days ago are cleaned up. If a member has not synched within this retention period, the service will detect it as an outdated member and prevent that member from synching.
It sounds like there is a way to safely trim this data, is there a way to trigger this for a shorter period? Our database syncs numerous times throughout the day, and it would be safe to trim at 1 day of retention data.