Monitoring Networked Applications With Incremental Quantile Estimation - Statistics > MethodologyReportar como inadecuado

Monitoring Networked Applications With Incremental Quantile Estimation - Statistics > Methodology - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

Abstract: Networked applications have software components that reside on differentcomputers. Email, for example, has database, processing, and user interfacecomponents that can be distributed across a network and shared by users indifferent locations or work groups. End-to-end performance and reliabilitymetrics describe the software quality experienced by these groups of users,taking into account all the software components in the pipeline. Each userproduces only some of the data needed to understand the quality of theapplication for the group, so group performance metrics are obtained bycombining summary statistics that each end computer periodically andautomatically sends to a central server. The group quality metrics usuallyfocus on medians and tail quantiles rather than on averages. Distributedquantile estimation is challenging, though, especially when passing largeamounts of data around the network solely to compute quality metrics isundesirable. This paper describes an Incremental Quantile IQ estimationmethod that is designed for performance monitoring at arbitrary levels ofnetwork aggregation and time resolution when only a limited amount of data canbe transferred. Applications to both real and simulated data are provided.

Autor: John M. Chambers, David A. James, Diane Lambert, Scott Vander Wiel


Documentos relacionados