[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: dissection of search latency

You wrote:
> Fascinating. Assuming that both backends were built using the same version
> of Berkeley DB, I'd guess the slowdown in back-bdb is due to transaction
> management. Except, the search code is not transaction-protected, so it's
That doesn't matter much. If you have outstanding transactions and you run your
search with a null txn, bdb still has to check against the locks of the outstanding
transactions (and cursors with null txn).

If at first you don't succeed, destroy all evidence that you tried.