A caching system with object sharing
share
We consider a public content caching system that is shared by a number of proxies. The cache could be located in an edge-cloud datacenter and the the proxies could each serve a large population of mobile end-users. The proxies operate their own LRU-list of a certain capacity in the shared cache. The length of objects appearing in plural LRU-lists is divided among them. We provide a "working set" approximation to quickly approximate the cache-hit probabilities under object sharing. We also discuss an approach to sharing cache I/O based on token bucket mechanisms. Also, why and how a proxy may issue mock requests to exploit the shared cache is discussed.
READ FULL TEXT