On 02/17/2015 08:21 AM, Yuri Voinov wrote:
squidGuard does not support the Squid feature 'concurrency' for url_rewrite_children. ufdbGuard does.
With concurrency, latency goes down and the number of processes can also be reduced.
The lack of concurrency is main disadvantage of squidGuard. Old process-based model is not acceptable on modern OS'es.
BTW, can I configure ufdbGuard children like this:
url_rewrite_children 1 startup=1 idle=1 concurrency=100
Yes you can, but it is not recommended :-( since throughput is higher with
url_rewrite_children 50 startup=5 idle=2 concurrency=3
When concurrency was built into the ufdbGuard redirector (named ufdbgclient)
it was done to increase the flow of requests:
Squid <--pipe--> ufdbgclient <--UNIX-socket--> ufdbguardd daemon
To keep ufdbgclient busy, there must always be a next request in the pipe and
and concurrency of 2 or 3 does that. It is not optimised for having
100 requests in the pipe.
Since ufdbgclient is leightweight, there is no problem with having 30 or more
of them and therefore ufdbgclient was never optimised for the use case of
url_rewrite_children 1 startup=1 idle=1 concurrency=100
Marcus
_______________________________________________
squid-users mailing list
squid-users@xxxxxxxxxxxxxxxxxxxxx
http://lists.squid-cache.org/listinfo/squid-users
