Re: [squid-users] histograms

Histogram statistics are done by collecting the samples into a set of
bins, one for each of the sampled range (for example service time). The
bins are usually distributed logaritmically over the covered measurement
range.

The mathematics shown are:

  bin_index/left_border hit_count (hit_count/bin_size)

The rightmost value is a normalised value showing hits per area covered
by the bin. Makes more sense further down in the histogram where the
bins are much larger.

How to actually read anything useful from these histograms is another
business, and depends very much what you are looking for. But it is
probably a good idea to start looking at the left_border (the value
after the /) and normalized hit_count (the rightmost value).

--
Henrik Nordstrom
Squid Hacker
Pranav Desai wrote:
> 
> hi!
> can anyone explains how to read this histogram-
> 
> client_http.all_svc_time histogram:
>           0/0.000000    87      1568.494232
>          13/1.017352    13      116.178383
>          20/1.943709    6       36.746840
>                 .       .           .
>                 .       .           .
>                 .       .           .
> 
> -pranav