I'm looking to peer Squid caches at two sites seperated by a high-speed, low latency link. Users connect to their nearest Squid and fail over to the far side if necessary.
I'm not sure of the best peering mechanism to use in this situation. Both Squids have fairly slow upstream links, so getting a hit is a priority. For cache hits, disk and CPU is more likely to be a bottleneck than the network, so it would be nice to have cached objects evenly distributed and not duplicated. Should I use ICP, HTCP, or Cache Digests? Regards Gerard
