ABSTRACT

Introduction ......................................................................................................... 316 Data Stream Processing ............................................................................. 316 Stream-Based Joins .................................................................................... 316 Application Scenario ................................................................................. 317

Existing Approaches and Problem Definition ................................................ 318 Proposed Solution ............................................................................................... 321

Execution Architecture .............................................................................. 321 Algorithm .................................................................................................... 323 Asymptotic Runtime Analysis ................................................................. 324 Cost Model .................................................................................................. 325

Memory Cost .................................................................................. 325 Processing Cost .............................................................................. 326

Analysis of w with Respect to its Related Components....................... 327 Effect of the Size of the Master Data on w ................................. 328 Effect of the Hash Table Size on w .............................................. 328 Effect of the Disk Buffer Size on w .............................................. 329

Tuning .......................................................................................................... 329 Tests with Locality of Disk Access ....................................................................330 Experiments .........................................................................................................333