Jakarta Lucene MultiSearcher Question

[asteiger]

This question is directed towards those familiar with using the Lucene
text indexer.

In my situation, there will be one, most likely, fairly large index,
and over time a trickle of smaller indexes being created that could
eventually number into the hundreds. Does using MultiSearcher to search
against all these separate indexes impose a performance hit as compared
to merging the smaller indexes into the original larger one? How long
could a typical index merge take, just arbitrarily?

Thanks,
Ashley