Next to guaranteeing serialisability, OLAP systems aim foremost at improving query
response time. The idea is to introduce freshness of data as a new quality-of-service
parameter for transaction processing. This should allow for explicitly trading freshness
of data accessed for query performance. OLAP clients specify a lower bound
of freshness for the data accessed by queries of the current transaction t, denoted
by ft. The freshness limit is an additional constraint for query routing.
In the following, we introduce a freshness-of-data-driven approach to replication
management in a database cluster: Freshness-aware scheduling (FAS). FAS comprises
replication management and mechanisms of multiversion concurrency control.
The notion of freshness of data is crucial in the context of FAS. We will presently
discuss freshness metrics, before introducing FAS in more detail.
244 R?¶hm
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of
Idea Group Inc. is prohibited.
Freshness.of.Data
Freshness measures are closely related to the notion of coherency for which several
ideas have been proposed in the literature (e.g., Alonso, Blott, Fe??ler, & Schek,
1997; Pacitti & Simon, 2000). Recently, those concepts have been revived under
the terms freshness of data (R?¶hm, B?¶hm, Schek, & Schuldt, 2002) or currency
(Guo, Larson, Ramakrishnan, & Goldstein, 2004).
Pages:
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461