Timer-based pre-fetching for increasing hazard rates

Caching plays a crucial role in today's networks: keeping popular content close to users reduces latency. Timer-based caching policies (TTL) have long been used to deal with bursts of requests, and their properties are well understood. However, in some scenarios the traffic is more regular. In this work, we define a dual of the TTL policy, dubbed Timerbased Pre-fetching, particularly well suited for these scenarios. We also show how the optimal timers can be computed for a large class of request processes.