Supporting block I/O contexts
Posted Jun 22, 2012 17:32 UTC (Fri) by giraffedata
Parent article: Supporting block I/O contexts
This sounds like a case of overgeneralization. There's no way an abstract, undefined concept of "context" of a write (or is it of the data written?) can be useful. They should just have made it (and called it) expiration group.
Separately, an expected lifetime value would help with the placement goals that apparently underly this feature.
to post comments)