The client I just mentioned wouldn??™t have a reason to mix the SQL data with the home
directories, and therefore, wouldn??™t have had to deal with an angry CEO. Mount points
provide granularity for controlling clustered disk resources previously unavailable to the
SQL Server DBA.
CHAPTER 7 n CLUSTERING 180
CLUSTER-ABLE VS. CLUSTER-AWARE
Simply clustering two or more servers together doesn??™t guarantee that when one node fails, everything
will resume on a second node. Some applications and services are designed to detect a failover event
generated by the cluster service, while others are not. Applications or services that have no direct interface
with information presented by the cluster service must use alternate means of responding to a
failure.
To detect cluster events such as a failover, applications and services must be cluster-aware. They
don??™t necessarily appear as cluster resources that you can configure for failover, but they have the ability
to listen to the cluster service and react accordingly, if programmed properly. They respond to
shutdown requests from the cluster service first and foremost, because a failure event could be caused
by something that allowed the cluster service to shut down programs cleanly before taking itself offline.
Pages:
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351