Nguyen, Khanh, INFOT wrote:
Errol, If you are familiar with scripting language like Perl, awk, you can quickly process the access.log (not the cache.log) for MISS entry, list of URLs that are MISS, how many of them, Approximate how long the object stay in cache (a hit then a miss on the same url) Are they including the query string.. Those data would be handful for you to figure out a policy to improve the hit rate and bandwidth saving. Alternative to scripting, you can try grep utility of unix, but you will have lot more data to analyze by eyes. For example: 'grep MISS access.log > output.txt' to get all miss entries and output to file output.txt Not sure if there is any package, utility out there to do this kind of analysis. Maybe a nice utility tool to analyze the log and give out some suggestions to tune it for higher hit rate. One thing i found useful when tunning is to go after the objects that being accessed the most. Maybe a very handful of objects that result low hit rate, if you can correct those, your goal is almost done. Khanh
As part of it's output Scalar (http://scalar.risk.az/) quantifies hit/miss statistics for file extensions, content type and Object sizes. Very useful for tuning.
Chris
