On the effects of content compression on Web cache performance

Compression is widely used to reduce data transmission costs, but its impact on the performance of network entities such as Web caches has not yet been studied. We evaluate the impact of compression on cache performance when the cache compresses content and the client decompresses it. We further determine the performance bounds by benchmarking the best-case (server compression-client decompression) and the worst-case (cache compression and decompression) scenarios. Our results show that caching compressed content significantly improves the cache's hit ratio, reduces user latency and increases cache throughput. With a compression ratio of 1.69:1, we see a 43% increase in hit ratio, a 12% decrease in average end-to-end latency and a 29% latency reduction for cache hits. We also find that cache hit ratio is critical to average end-to-end latency. Even in the worst-case, the cost of compressing and decompressing content is outweighed by the benefits at a reasonable compression ratio of 1.58:1.