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

Re: dissection of search latency

You wrote:
> The entry caching of back-ldbm proved extremely effective. There's no I/O
> overhead in back-ldbm once all entries are cached and the cache hit is
> almost 100 times faster than disk access. Thus, it may be desirable
> to have efficient entry caching in back-bdb as well, even though
> further investigation should be necessary.
I've done some tests on the cache mutex to see if it was heavily contended, but
this is fortunately not the case in normal conditions. So, one big cache
mutex is not a problem.

However, I may suggest an incrementing backoff time in the goto construction, as this
really bogs down your server under heavy load. (stable threading assumed...)

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