Professional SQL Server™ 2005 Programming

20.1. Replication Basics

Replication is like a big puzzle — made up of many pieces in order to form a complete unit. We have topology considerations (publisher, subscriber, and distributor) as well as publication models (merge, transactional, snapshot). Before you get to deciding on those, there are several things to take into account.

20.1.1. Considerations When Planning for Replication

There are a number of things to take into account when thinking about the topology and replication methods available. These should be part of an assessment you make at design time to determine what forms of replication should even be considered for your application. Among these are:

• Autonomy

• Latency

• Data consistency

Let's take a quick look at each of these.

20.1.1.1. Autonomy

Autonomy is all about how much a replication instance is able to run as its own thing. What data needs to be replicated and at what frequency? For example, you could be supporting a sales application where each site keeps separate customer records. You would want to have these replicated to a central database for reporting and, perhaps, such other things as automatic stock replacement. Each site is highly autonomous (they really don't care whether the central database gets its data or not; they can still continue to make sales based on the data they have on-site). Indeed, even the central database, while dependent, is probably not in a catastrophic situation if it misses data from a site for a day (depends how you're using ...