Next: cache document timeout Up: trace-driven analysis Previous: trace-driven analysis

requests per minute with caches

Figure 4 shows the number of document requests per minute to the NCSA server cluster; figure 5 divides them according to the source zone of the request using the zones in figure 3.

Figure 6 shows the the number of document requests per minute to the NCSA server for zone 8 given a local cache with a 4096 second timeout. The top line in the graph reflects requests for documents in the local cache, which include those requested for the first time from the central server as they are placed in the cache. The bottom line tracks this latter subset of queries, the set of documents that require transmission from the central server. The difference between these two lines illustrates the savings offered by the cache. Cache behavior for the other zones appears similar; we do not include their graphs here.


kc@
Thu Sep 15 22:53:05 PDT 1994