What is Cobalt InstaCache Clustering, and whats it good for?
InstaCache Clustering works as follows: all client traffic is directed to one CacheQube, the Cluster Master. This can occur transparently or through traditional proxy methods, just as if the Cluster Master were a standalone CacheQube. However, with Clustering, the Cluster Master does not actually service individual HTTP requests itself — instead, it passes off each request to one of several configured Cluster Slave CacheQubes. The Slave CacheQubes take care of retrieving objects from origin servers and then caching these objects. Much of the computational and I/O load of web caching is the disk I/O involved with storing and retrieving cached documents. The Master avoids this — only the slaves deal with storing documents. The master balances the storage load among the slaves. Putting this all together, a given HTTP request is handled as follows: 1 master: get the request from a client (transparently or non-transparently) 2 master: based on the request’s URL, pick a slave CacheQube to
