Improved byte caching techniques

Abstract

Byte caching (also known as redundancy elimination) is a commonly used technique in practice to compress data streams in data networks. The congestion of a bottleneck link can be significantly reduced when byte caching is implemented in both ends of the link. We propose four techniques to improve the original byte caching method. Two of these techniques offer better fingerprint sampling for caching and the other two offer better management of fingerprints to increase the likelihood of matching. The implementations of these techniques improve the compression ratio of the original byte caching method by 10-20%, which is a significant boost to the efficiency of byte caching methods. Proposed techniques can be implemented with minimal to moderate overhead compared with the original byte caching method.