Analyzing the Traffic Characteristics for Evaluating the Performance of Web Caching

Analyzing the Traffic Characteristics for Evaluating the Performance of Web Caching

G.P. Sajeev (Government Engineering College, India) and M.P. Sebastian (National Institute of Technology, India)
DOI: 10.4018/978-1-60960-523-0.ch012


Web cache systems enhance Web services by reducing the client side latency. To deploy an effective Web cache, study about traffic characteristics is indispensable. Various reported results show the evidences of long range dependence (LRD) in the data stream and rank distribution of the documents in Web traffic. This chapter analyzes Web cache traffic properties such as LRD and rank distribution based on the traces collected from NLANR (National Laboratory of Applied Network Research) cache servers. Traces are processed to investigate the performance of Web cache servers and traffic patterns. Statistical tools are utilized to measure the strengths of the LRD and popularity. The Hurst parameter, which is a measure of the LRD is estimated using various statistical methods. It is observed that presence of LRD in the trace is feeble and has practically no influence on the Web cache performance.
Chapter Preview


Web caching is emerged as a technique to leverage the quality of Web services (Barish & Obraczka, 2000). A proxy cache deployed near to a client serves the Web documents locally as shown in Figure 1. This increases the client side download speed and reduces the outborn traffic. An efficient proxy-cache server can deliver most of the requests from its local cache. The performance of such a service depends on the cache architecture, admission policy, replacement method and cache consistency. Handling the input traffic dynamics is the key aspect in a cache server’s success.

Figure 1.

A proxy-cache server system


In spite of the evolution of new caching schemes and algorithms, the performance of cache services never reached to the expected levels, as can be seen from Table 1, 2 and 3. Suitable setting of the cache performance parameters can provide a solution to this problem. Hit ratio, byte-hit ratio, client side latency and network load reduction (server side and network) are some important cache performance parameters. These parameters can be estimated by analyzing the input traffic stream and Web proxy traces.

Table 1.
Traces Used
TraceDateNumber of requestsHit RatioZipf slope
NLANR-pa1 'st Sep 20072,80,06236%0.72
NLANR-pa1 'st Oct 20072,10,73432%0.76
NLANR-bo210'th Aug 20072,43,35626%0.69
NLANR-sj9'th Jan 20075,44,35620%0.80
Internet Traffic Archive23 'rd Jan 200528,33848%0.72

Complete Chapter List

Search this Book: