Hi, How do I measure the non-proxied service time? (Sorry if the question sounds stupid. I appreciate if you lead me to where to find out.) On Sun, Mar 27, 2011 at 7:02 PM, Amos Jeffries <squid3@xxxxxxxxxxxxx> wrote: > On Sun, 27 Mar 2011 06:33:18 -0400, January Sharp wrote: >> >> Hi, >> >> We installed Squid 3.1.10 three days ago, and lots of clients are >> using it. Being a complete newbie to this, may I ask a question. >> >> What is best statistic to use to find out the general, overall >> improvement of our clients' web surfing with vs. without squid. That >> is, when I report to my management, "Our users are able to surf X% >> faster since we installed squid". >> >> Is this X available or computable from the information in the General >> Runtime Information page of the Cache Mgr? Or somewhere else in Cache >> Mgr? >> >> I am guessing it is 1 - (Cache Misses Median Service Times / HTTP >> Requests (All) Median Service Times). Is this correct? > > Any of the metrics which provide both HIT and MISS details can be used as > performance gain measures. > > Speed improvements are complex. For absolute statements like that you will > have to combine the HIT service time with a non-proxied service time. > Mediated in some calculation against the request HIT ratio. > The results are rarely what you expect. Adding a non-caching proxy usually > results in speed LOSS from the overheads, but not always. > > The other metric used are Byte Hit ratios/totals. Combined with your > upstream bandwidth costs that measures a direct $$ figure in the budget. > > Amos > >
