[prev in list] [next in list] [prev in thread] [next in thread] 

List:       slide-dev
Subject:    Re: Perfomance observations
From:       Stefan_Lützkendorf <luetzkendorf () apache ! org>
Date:       2005-02-28 11:10:19
Message-ID: 4222FC1B.3000209 () apache ! org
[Download RAW message or body]

Ok, the picture doesn't get it to the list. The picture shows something like:

  time per PROPFIND request
  |
  |                                           +
  |                                         +
  |                                       +                       + with lock-cache-size 100
  |                                     +                         * with lock-cache-size 10000
  |                                   +
  |                                 +
  |                               +                       *
  |                             +                    *
  |                           +                 *
  |                         +              *
  |                       +           *
  |                     +        *
  |                   +     *
  |                 +  *
  |               +
  |          +
  |     +
  |+
  +----------------------------------------------------------------------> #docs in collection
                  100                                                 400

Above point "100 docs in collection" the time per request starts to grow faster.

Stefan Lützkendorf wrote:

> Hello folks,
> 
> while investigating slides performance with large collections I observed
> that the default store parameter lock-cache-size is to small for the 
> txfile store.
> 
> The cache size for nodes, revistion descriptors, etc is per default 
> 10000, but
> for locks it is 100. That means if you have all you nead for a request 
> in the cache,
> but not the locks you have to reread the xml resource descriptor anyhow.
> 
> I attach a picture that illustrates what i mean.
> 
> Is there any point against setting the default lock-cache-size to the 
> same as
> the other caches?
> The only point I see is that this cache caches vectors that almost all 
> empty.
> I'll modify this, so that empty vectors are not cached.
> 
> 
> Regards, Stefan
> 
> 
> ------------------------------------------------------------------------
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: slide-dev-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: slide-dev-help@jakarta.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: slide-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: slide-dev-help@jakarta.apache.org

[prev in list] [next in list] [prev in thread] [next in thread]