??? Hardware: A hardware disaster is really the only scenario in which clustering has
any appreciable impact. If any physical node in the cluster fails, the only data lost
will be the current transaction.
CHAPTER 7 n CLUSTERING 192
??? Media: Clustering provides no protection from media disasters, because all SQL
Server data is kept on a shared device. If that device fails, all data is lost. If the
media for the individual server (holding the operating system) is damaged, the
ability to fail over to another node will protect you to a point. If the media storing
the quorum is damaged, the clustering will fail altogether.
??? Process: There are some scenarios in which clustering can protect against process
errors. If the error is a long-running, incorrect transaction (such as issuing a DELETE
without a WHERE clause), initiating a manual failover will cause the incorrect transaction
to roll back. Of course, hard-booting a server would accomplish the same
thing, but in a cluster environment, the users would have only a brief hiccup in
availability.
??? User: As with process errors, if you can catch the user error in time (which implies
you know what was done when, which is rare), you could use a manual failover to
roll back the mistake while maintaining availability.
Pages:
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373