Currently FileBackedStorageEngine sizes each new chunk as twice the size of it's predecessor. This means shrinking currently requires freeing half of all stored entries. We should consider changing this to some scheme where chunk sizes are capped so that shrink can have a less drastic implementation.
Currently FileBackedStorageEngine sizes each new chunk as twice the size of it's predecessor. This means shrinking currently requires freeing half of all stored entries. We should consider changing this to some scheme where chunk sizes are capped so that shrink can have a less drastic implementation.