Using latency-recency profiles for data delivery on the web

  • Authors:

  • Laura Bright;Louiqa Raschid

  • Affiliations:

  • University of Maryland, College Park, MD;University of Maryland, College Park, MD

  • Venue:
  • VLDB '02 Proceedings of the 28th international conference on Very Large Data Bases
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

An important challenge to web technologies such as proxy caching, web portals, and application servers is keeping cached data up-to-date. Clients may have different preferences for the latency and recency of their data. Some prefer the most recent data, others will accept stale cached data that can be delivered quickly. Existing approaches to maintaining cache consistency do not consider this diversity and may increase the latency of requests, consume excessive bandwidth, or both. Further, this overhead may be unnecessary in cases where clients will tolerate stale data that can be delivered quickly. This paper introduces latency-recency profiles, a set of parameters that allow clients to express preferences for their different applications. A cache or portal uses profiles to determine whether to deliver a cached object to the client or to download a fresh object from a remote server. We present an architecture for profiles that is both scalable and straightforward to implement at a cache. Experimental results using both synthetic and trace data show that profiles can reduce latency and bandwidth consumption compared to existing approaches, while still delivering fresh data in many cases. When there is insufficient bandwidth to answer all requests at once, profiles significantly reduce latencies for all clients.