> -----Original Message----- > From: Raj [mailto:sunfire2005@xxxxxxxxx] > Sent: Friday, February 17, 2006 2:46 PM > To: Chris Robertson > Cc: squid-users@xxxxxxxxxxxxxxx > Subject: Re: parent cache information > > > Chris, > > Once again thanks heaps. You were absolutely spot on. We have total 4 > proxies (2 child & 2 parent proxies). > > Server A & Server C - parent proxies (2nd tier) > Server B & Server D - Child proxies (1st tier) > > 1st Tier uses NTLM authentication via the the Samba WINBIND process. > 2nd Tier is located in the DMZ with no authentication required. > > This is the main reason we are using 1st tier and 2nd tier proxies. > For this type of setup could you please recommend whether to configure > both proxy's to cache or just 2nd tier proxies as cache and 1st tiers > as proxy only. Basically I want to achieve better performance than > what we have now. At the moment as explained to you before both 1st > tier and 2nd tier are caching. > > Once again thanks a million. > > Regards. > Well, I think you will see your best improvement by recompiling and including aufs. In any case, for a different reason I have a child-parent hierarchy on a single LAN segment. My cache_peer line on one child (proxy1) is: cache_peer proxy2 sibling 8080 3130 proxy-only no-digest cache_peer proxy3 parent 8080 3130 proxy-only round-robin no-digest cache_peer proxy3 parent 8081 3131 proxy-only round-robin no-digest Both my request hit and byte hit ratio on the child proxy are low (but non-zero) numbers. Perhaps that indicates that only cached requests fetched from the parent proxy are not cached on the child, vs. all requests. Then again, due to other quirks with my setup that metric may not be indicative of anything. As for myself, I can perceive no difference between surfing with or without the proxy. Anecdotal evidence at best. Chris
