The instance is slow after a cache flush.

This can also be translated in high response time, coupled with default semaphore exhaustion on all nodes: HTTP 429 errors (Too many requests) 

A full system cache flush means that all transactions have to pull their data from the database, and this makes them slower than when the cache has been rebuilt.

This can be initiated this with the '/cache.do' transaction.

Full cache flushes can be triggered also by updates to the 'sys_properties' table, as well as by deploying update sets.

We advise not performing this action during core business hours if at all possible.