Dynamic Server Selection in the Internet

this paper we report on techniques for finding good service providers without a priori knowledge of server location or network topology. We only assume that the client has been provided with a list of addresses of servers that provide the required service. We focus on the problem of dynamically selecting a server for replicated documents, but most of our results are applicable to replicated services in general. We consider the use of two principal metrics for measuring distance in the Internet: hops, and round-trip latency. We show that these two metrics yield very different results in practice. Surprisingly, we show data indicating that the number of hops between two hosts in the Internet is not strongly correlated to round-trip latency. Thus, the distance in hops between two hosts is not necessarily a good predictor of the expected latency of a document transfer. Previous work on server selection has emphasized static selection, relying on a previous server discovery step [5], or random selection [2]. Static selection work has used hops as the distance metric, since this metric varies least over long periods. We show in this work that the extra cost at runtime incurred by dynamic latency measurement, as compared to prior static knowledge of hop distances, is well justified based on the resulting improved performance. In addition we show, not surprisingly, that selection based on dynamic latency measurement is preferable to random selection. The difference between the distribution of hops and latencies is fundamental enough to suggest differences in algorithms for server replication. We show that conclusions drawn about service replication based on the distribution of hops need to be revised when the distribution of latencies is considered instead. 2 Hops Vs. Latenc