That is another benefit of having a caching proxy server. Regardless of whether there are hardware limitations on memory and hard drive space, proxies can be set-up as peers. And this is not limited to just two (2) peers. You can have many peers set-up to help each other out with the caching. Fetching from local cache is faster than fetching the same information from the web, which definitely helps speed-up browsing.
It doesn't take rocket science to set this up. The same procedure in setting up a squid caching proxy server is followed. The configuration file is the pretty much the same, only tweak it a little to let squid know the peer exists. This part of the previous squid.conf is very important.
icp_port 3130 icp_query_timeout 5000
It is required for peer caches to be able to query each other. The directive icp_port 3130 tells squid to listen to port 3130 for peer queries. And icp_query_timeout of 5000 means the timeout of a peer query is 5 seconds, ample enough to wait for timeout between proxies. That is what I have in my configuration, test it out to see which applies best in your environment.
The other directive that is required for a peer cache proxy is cache_peer. It lets squid know who the peers are and which port it listens to. Further information about this directive and its options are on the squid documentation. The code to use is below:
cache_peer [peer_proxy_name or ip_address] http_port icp_port options
Note that in the above syntax, http_port and icp_port both refer to the ports used on the peer, not on the local proxy server. It helps to simplify things by using the same ports for peer caches.
In order to see if the peer caches really query each other, you may wish to turn on access_logging. Checking the squid.conf.default file you can see that the original directive for this is:
access_log c:/squid/var/logs/access.log squid
Just replace the line "access_log none" with the above code. Then, re-configure squid to reflect the changes.
After browsing for a bit, check the log file and see if the IP address (or hostname) of the peer cache proxy appears on the log. That should be a good indication that the proxies are now helping each other out.
On this part, it is easy to miss the portion where you configure the browser. I have had many experiences of trying to troubleshoot the proxy (and wasting precious time in the process) only to find out the browser was never configured to use the proxy in the first place. Lessons learned here, so the first question I always ask is: Have you configured the browser to use the proxy?
