More exactly, a replication topology is more like a distributed database.
Any other uses of replication (and there are a lot of creative uses) are for secondary
purposes only.
nNote SQL Server 2005 introduces a new type of replication: peer-to-peer. Normally, replication topologies
follow a publisher/distributor/subscriber model. In the peer-to-peer model, every server involved is essentially
both a publisher and a subscriber. There is, of course, a latency issue as the individual nodes sync
themselves, but this could be a possibility to consider as a mitigation technique. There??™s also a point of
diminishing return by adding too many nodes??”the replication overhead becomes too complicated and
costly. I??™ve spent a lot of time with replication, and until I put peer-to-peer through the wringer so to speak,
I??™m not going to endorse it.
CHAPTER 6 n MAINTAINING A WARM STANDBY SERVER VIA LOG SHIPPING 142
Replication also incurs significant overhead, both in resources required and
administration and maintenance. It??™s a relatively complex technology, meaning there
are more points of possible failure than with log shipping.
Pages:
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287