[leveldb] bloom filter for avoiding disk reads for missing keys?

['Dan Rodney' via leveldb]

We are considering using LevelDb to store a set of strings (where the
values are all single byte dummy entries); we will call Get to find out
whether a given string is in the set.

We expect that most of the time, the keys we're passing won't actually
match anything in the set; we'd like to minimize the chances of reading
from disk in those cases if possible.

It seems like the Bloom filter support built into LevelDb may help with
this, but it wasn't entirely clear from the documentation, which instead
talks about a single call to Get causing multiple disk reads without it.
Our assumption is that given that our values are only one byte long, that
wouldn't happen... we just want to avoid doing any disk seeks for keys that
aren't there. Will the bloom filter support in LevelDb help with that?

Thanks,

Dan Rodney