Application-level differentiated multimedia Web services usingquality aware transcoding

摘要

The ability of a Web service to provide low-latency access to itsncontent is constrained by available network bandwidth. While providingndifferentiated quality of service (QoS) is typically enforced throughnnetwork mechanisms, in this paper we introduce a robust mechanism fornmanaging network resources using application-specific characteristics ofnWeb services. We use transcoding to allow Web servers to customize thensize of objects constituting a Web page, and hence the bandwidthnconsumed by that page, by dynamically varying the size of multimedianobjects on a per-client basis. We leverage our earlier work onncharacterizing quality versus size tradeoffs in transcoding JPEG imagesnto supply more information for determining the quality and size of thenobject to transmit. We evaluate the performance benefits ofnincorporating this information in a series of bandwidth managementnpolicies using realistic workloads and access scenarios to drive ournsystem. The principal contribution of this paper is the demonstrationnthat it is possible to use informed transcoding techniques to providendifferentiated service and to dynamically allocate available bandwidthnamong different client classes, while delivering good quality ofninformation content for all clients. We also show that it is possible toncustomize multimedia objects to the highly variable network conditionsnexperienced by mobile clients in order to provide acceptable quality andnlatency depending on the networks used in accessing the service. We shownthat policies that aggressively transcode the larger images can producenimages with quality factor values that closely follow the untranscodednbase case while still saving as much as 150 kB. A transcoding policynthat has knowledge of the characteristics of the link to the client cannavoid as many as 40% of (unnecessary) transcodings