That means less data
to back up, which means a faster backup.
As an example, assume you have a sizeable database (around 500GB) that you need
to back up as quickly as possible. Your primary reason for backing up is data protection,
not data availability. You can separate all of the nonclustered indexes onto a separate filegroup
so you can isolate it in your primary backup plan.
The resulting design could consist of three filegroups: one named PRIMARY (holding
only the .mdf file), another for data (.ndf files holding data tables only), and one for
indexes (.ndf files holding only nonclustered indexes). Your backup scheme would look
something like this:
??? PRIMARY filegroup (Sunday at 10 p.m.)
??? Data filegroup (Monday through Friday at 10 p.m.)
??? Index filegroup (just once)
??? Log backup (Monday through Friday, every four hours)
You aren??™t continually backing up index data; that filegroup is only backed up once
so that you have a complete set of backups. You could even shortcut that step if you
want to??”more on that shortly. Yes, it will take longer until the database functions effectively,
but what you care about is a short backup time and protecting the data.
Pages:
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257