Now the following query Qx has to be routed:
S E L E C T c o u n t ( * ) F R O M O r d e r s , L i n e I t e m W H E R E . . .
It accesses the two relations Orders and LineItem. The intersection with
the approximated cache state of node N1 is{LineItem}, which is a caching
benefit of 1; the intersection with the cache state of node N2 is {Orders,
LineItem}, corresponding to a caching benefit of 2. Hence, CAF would
route query Qx to node N2, because it expects there a higher caching
benefit.
242 R?¶hm
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of
Idea Group Inc. is prohibited.
This approach relies on a very rough approximation of queried data. It does not take
into account which portions of the relations are actually accessed by a query. These
portions are described by each query predicate. However, quantifying the exact
overlap of the sets of tuples specified by two different predicates having common
attributes is difficult. In R?¶hm et al. (2001), an interesting refinement based on bit
strings has been proposed. The authors proposed to further approximate the set of
tuples specified by a predicate using bit string signatures. By doing so, one can reduce
the calculation of benefits to bit string comparisons. This can be done efficiently.
Freshness-Aware. Scheduling
The previous considerations have concentrated on a query-dominant environment.
Pages:
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457