In particular, we want to be able to easily plug together and to expand the cluster
using standard hardware and software components only. This results in a highly
scalable system architecture.
We concentrate on central architectural issues and performance aspects of database
clusters for usage in a decision support scenario. The objective is to develop a basic
infrastructure for interactive decision support systems that are capable of analysing
up-to-date data and that can give guarantees on how outdated data accessed might
be. To be able to do so, we need different versions of data in the cluster, which we
achieve by replicating data throughout the cluster. Replication also helps to avoid
expensive distributed joins over huge amounts of data; as always several nodes can
2 2 R?¶hm
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of
Idea Group Inc. is prohibited.
evaluate an OLAP query. We will discuss query routing strategies for an optimal
workload distribution of long running and I/O intensive OLAP queries over encapsulated
standard components of a database cluster.
Furthermore, we explicitly allow that not all cluster nodes are up-to-date all the
time. This is reflected by the notion of freshness of data, which is a measure for
the deviation of a certain component as compared to an up-to-date component.
Pages:
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437