On 10/5/2012 11:08 AM, Stone, Joel wrote:
OK I'll bite - could you please share what some of the major gotcha's were with the home grown replication systems that didn't work so well?
1) The 'two of everything' problem. You have two database schemas, two
development teams, two maintenance schedules, two upgrade schedules, two
management teams with different priorities. From now on, whenever one
side wants to change, the other side will need to be consulted.
2) The 'separate but identical' problem. The actual mechanics of
synchronising two databases are not trivial. For example, what happens
when SQL Server is offline for backups, or Patch Tuesday or because the
breaker tripped in the server room? Things happen, one of the machines
is going to be unavailable when the other one wants it. This will force
you to create a process to verify that the databases are in sync. Which
probably means that both databases will need to be taken offline. Of
course, once you find an out of sync condition you'll need a process to