On 08/10/2018 09:51 AM, Cheadle, Edward wrote: > If I do split the rules up, does it increase the speed of processing the > rules? If by "processing" you mean applying configured rules to traffic at runtime, then the answer is "no". At runtime, Squid uses compiled rules, and compiled rule application/checking speed does not depend on where the rule were loaded/compiled from. > I was told it does, but in looking at information on the squid > site, in how rules are processed, I can’t find anything that would lead > me to believe that splitting out the rules into separate files would > speed anything up. If the deployment environment loads/reads few large files slower than it loads/reads many small files, then it is possible that using many small configuration files will speed up Squid configuration. Also, loading lots of ACL values/parameters from a file might be faster than loading lots of ACL values/parameters embedded in squid.conf (because it might be easier for Squid to find where each value starts/ends in an external parameter file), but I have not tested that theory, and, again, this theory only applies to Squid (re)configuration delay, not runtime processing/performance. > The reason this is an issue, is because on AWS we have a load balancer > and we spin up an instance and if it takes too long it tends to stop the > instance. What is the approximate total size of your Squid configuration, including all the external files that Squid has to load/parse while configuring itself? What is the current startup delay with a trivial/small rule set? What is the additional startup delay from parsing all those rules? Alex. _______________________________________________ squid-users mailing list squid-users@xxxxxxxxxxxxxxxxxxxxx http://lists.squid-cache.org/listinfo/squid-users
