Mailing-List: contact dev-help@trafficserver.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@trafficserver.apache.org
From: jplevyak <git@git.apache.org>
To: dev@trafficserver.apache.org
Reply-To: dev@trafficserver.apache.org
References: <git-pr-364-trafficserver@git.apache.org>
In-Reply-To: <git-pr-364-trafficserver@git.apache.org>
Subject: [GitHub] trafficserver pull request: TS-4053: Add regression tests
 for hit ...
Content-Type: text/plain
Message-Id: <20151205010036.7BFCAE08EF@git1-us-west.apache.org>
Date: Sat,  5 Dec 2015 01:00:36 +0000 (UTC)

Github user jplevyak commented on the pull request:

    https://github.com/apache/trafficserver/pull/364#issuecomment-162120619
  
    The upshot is that the two Ram caches use the correct amount of memory (within 2%) and that LRU works better for identically sized objects (because LRU is a very good proxy for hit rate and LRU has less memory overhead) and CLFUS works better for variable/mixed size objects (which is expected since that is what it is trying to do).  Note that for large caches, the cost of CLFUS for fixed size approaches zero as the overhead has less effect but the benefit for variable size objects increases.
    
    The regression tests at 1MB 16MB and 256MB, but with relatively small objects (16KB) so the results should be applicable to more popular production size RAM caches in the GB(s) range.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---