I've recently started testing out the viability of using no caching at
all on my 2 Squid proxies as I believe the checking of my cache_dir's is
possibly slowing the proxying down a whole lot. In switching to no
caching I've noticed that on my dual-processor servers Squid tends to
max out one of the processors. With this in mind and the fact that
Squid will, with a single thread, monitor one port, would it make sense
to have Apache provide round-robin access between this Squid and another
Squid running on a different port? Say, 3129 (instead of the default,
already running 3128?) This is simply to spread the load, on one
server, between two Squids, two threads, two processors? Is this
completely erroneous? Am I wrong in any of my thinking here?
Opinions welcome.
Peter Smith
Linux Systems Administrator
University of Texas Southwestern Medical Center at Dallas
214 648 3111
peter.smith@utsouthwestern.edu
Received on Mon Jan 14 2002 - 09:51:58 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:05:51 MST