Alternative path selection in resilient web infrastructure using performance dependencies
Journal of Web Engineering
Using non-random associations for predicting latency in WANs
WISE'05 Proceedings of the 6th international conference on Web Information Systems Engineering
| Hi-index | 0.00 |
Recent technological advances have enabled the deploymentof wide area applications against Internet accessiblesources. A performance challenge to applications in sucha setting is the unpredictable end-to-end latency of accessingthese sources. We use passive information gatheringmechanisms to learn end-to-end latency distributions andconstruct Latency Profiles (LPs). We hypothesize that agroup of clients, within an autonomous system (AS), thatare accessing a content server, in another AS, may be representedby (one or more) LPs. Related networking researchon IDMaps, points of congestion, and BGP routes supportsuch hypothesis. We develop aggregate LPs to provide coverageof groups (clusters) of client-server pairs. Using datagathered from a (limited) experiment we demonstrate thefeasibility of constructing LPs.