Saving Cash by Using Less Cache

Everyone loves a large caching tier in their multi-tier cloud-based web service because it both alleviates database load and provides lower request latencies. Even when load drops severely, administrators are reluctant to scale down their caching tier. This paper makes the case that (i) scaling down the caching tier is viable with re- spect to performance, and (ii) the savings are potentially huge; e.g., a 4x drop in load can result in 90% savings in the caching tier size.