average number of requests using same TCP connection = 1 / (probability of
fresh-connect each time when making a request)
set 0<x<1, y = 1+x+x^2+x^3+...,
==> y = 1 + x (1 +x+x^2+x^3+...) = 1 + xy ==> y = 1 /(1-x)
==> (1+x+x^2+x^3+... )= x/(1-x)
avg# should be L = SUM (n*Pn), n=1, 2, 3.... (n is number of requests in a
certain TCP connection, Pn is the presence probability of n)
set p as probability of fresh connect each time when making a new request,
so
L = 1p + 2p(1-p) + 3p(1-p)^2 + 4p(1-p)^3 + ...
L/p = 1 + 2(1-p) + 3(1-p)^2 + 4(1-p)^3 + ...
= 1 + (1-p) + (1-p)^2 + (1-p)^3 + ...
+ (1-p) + (1-p)^2 + (1-p)^3 + ...
+ (1-p)^2 + (1-p)^3 + ...
= 1(1 + (1-p) + (1-p)^2 + (1-p)^3 + ... )
+ (1-p)(1 + (1-p) + (1-p)^2 + (1-p)^3 + ... )
+((1-p)^2)(1 + (1-p) + (1-p)^2 + (1-p)^3 + ...)
+ ...
= (1 + (1-p) + (1-p)^2 + (1-p)^3 + ... )(1 + (1-p) + (1-p)^2 + (1-p)^3
+ ... )
= (1/(1-(1-p))) * (1/(1-(1-p)))
= 1/(p^2)
so L = 1/p
So, could we set url->fresh_connect as re-connect probability instead of
always no or yes to better simulate the real world that has some length of
TCP reuse?
thanks,
Fred
2012/2/15 Fred Huang <di...@gm...>
> hello all,
>
> I am requesting below feature enhancements:
>
> 1. I found curl-loader's -r option never work but FRESH_CONNECT=1 always
> re-open new TCP connections. however, the real world browsing almost reuses
> TCP connection. the average of TCP reuse I have seen from a F5 loader
> banlancer of a large enterprise is 3.4 requests over 1 TCP connection. so
> could you add a configuration line such as "TCP_REUSE_RATIO" to enable
> opening new TCP connections with a probability, just like FETCH_PROBABILITY?
> this can greatly get close to real world.
>
> 2. url_cycling - its config parse function is commented in code. so the
> behavior on a URL_SET constructed from a URL_TOKEN_FILE is a client IP
> always send one URL. could this become configurable in config so that each
> client IP can fetch url one by one from same URL_SET?
> URL_TEMPLATE=%s
> URL_TOKEN_FILE=sfile.urls
>
> thanks,
> Fred
>
|
