, 2000).
A database cluster consists of commodity PCs, each running an off-the-shelf commercial
database system (DBMS) as a transactional storage layer. For the sake of
simplicity let us assume that all cluster nodes are homogeneous, that is, they run
the same DBMS with the same database schema. Each node holds a full copy of the
database. We also refer to a database at the cluster nodes as a component DBMS. We
distinguish between one or more dedicated master nodes and n secondary nodes.
There is a coordination middleware (also referred to as coordinator) that administers
the cluster. It is responsible for scheduling, routing, and logging of the incoming
requests. Except for this purpose-built coordinator, the cluster consists of off-theshelf
hardware and software components. The coordination middleware comprises
an input queue, a scheduler with an input queue, a router, a refresher, and a logger
(cf. Figure 2).
Figure 2. System architecture details (R?¶hm, 2002)
updates query query query
InputQueue
master node node node 2 node n
State
Info Refresher Scheduler
Router
Logger
CDBMS0 CDBMS CDBMS CDBMSn
EOT
log entry
Coordination Middleware
db db db db refresh
log
Cluster of Databases
global
log
OLAP with a Database Cluster 2
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc.
Pages:
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442