Feature cache works great as a method to inspect all transformers after a workbench has ran, but causes a few issues for me.

1. it hurts performance to cache each and every transformer
2. I lose all the cached data if FME crashes or I close the workbench.

I am looking for advice on ways to cache or otherwise save output from only a few transformers. My only idea is to write features out to a file, and later bring back in with a writer. are there other transformers I may not be aware of that can save/recall data after restarting the workbench?