Check out the new USENIX Web site. next up previous
Next: Benchmark Up: Experiments Previous: Experimental Setup

Test Data

Here we report a performance evaluation using a synthetic workload based on the multi-version archive of WWW pages collected by the AT&T Internet Difference Engine (described above in Section 3.1). This archive reflected the actual evolution of the pages, although it did not contain copies of every version of every page: some pages were archived automatically once per day when changes were detected, while the majority were archived upon the explicit instruction of a user of the system.

Slightly over half of the pages had only one version archived; these reflected pages that were registered with the system but had either never changed or (more likely) were not archived automatically and had not been selected for subsequent archival by a user. We excluded these pages from the benchmark because no deltas were available. On the other hand, about 10% of the 380 pages had 10 or more versions archived, and several had 50 or more versions (the latter were all pages that were archived automatically).



Gaurav Banga
Tue Nov 12 20:47:38 EST 1996