Hello all, What is the recommended approach to perform load balancing and high availability between N squid servers? I have the following list of requirements to fullfil:
1) Manage N squid servers that share cache (as far as i understand is done using cache_peer). Desirable. 2) Availability: if any of the N servers fails the clients are redirected to the rest N-1. Prefferable. 3) Scalability: The load is distributed (round-roubin or some other algorithm) between N servers, if a new server is added (N + 1) new clients will be able to use it reducing the load on the rest N. Prefferable. 4) I need to be able to identify client IP addresses on the squid side and/or perform Squid authentication. The client IP and User Name are later to be passed to the ICAP server to scan the HTTP(S) request/response contents using icap_send_client_ip and icap_send_client_username. Very important requirement. 5) I need to support both HTTP and HTTPS connections with support of selective SSL Bump. I.e. for some web sites I do not want to look inside SSL so that original site's certificates are used for encryption. Very important requirement too. I know that strictly for HTTP I could use HAProxy with Forward-For or something similar, but the 5th requirement is very important and I could not find a way to handle SSL with HAProxy properly. The only idea that comes to my mind is to use some form of round-robin load balancing on the level of DNS, but it has its own drawbacks (should be able to check availability of my N servers + not real balancing, more like distribution). Any help/thoughts are appreciated. Thank you! Best regards, Rafael Akchurin Diladele B.V. _______________________________________________ squid-users mailing list squid-users@lists.squid-cache.org http://lists.squid-cache.org/listinfo/squid-users
