Server Size:
With network requirements behind us, for data replication, data from distributed systems needs to be moved to central server. Central server holds duplicate data (subscriber in transactional replication or secondary server in case of log shipping and mirror in Database Mirroring) with main data at distributor server. Central server holds data from 350 distributor servers.
Transactional Replication:
Publisher server contains 1 DB instance that has 1 log reader agent, 1 distributor agent job, 1 cleanup job and log shipping monitoring jobs. Subscriber server would hold 350 subscriptions and thus would be very active compared to distributor server.
Log Shipping:
Log shipping should have 350 jobs for copy and 350 jobs for restoration. As number of jobs at central server are much higher than distributor server, server sizing at central server is very important.
Other points:
In log shipping secondary server could only be in read only mode (standby mode), there is not data modification possible.
In replication data modifications on secondary server is possible though that may lead to data / metadata mismatches between publisher and subscriber server.
Log Shipping and Transactional replication are both out of box solutions and are supported by Product support team.
Before designing data replication solution, multiple tests are to be carried out with different data replication solution. Main goals for each test case is
- Server Sizing both at Central server and Distributor server.
- Underlying Network capacity between Central and distributor server.
This is brief post for completing other parameters, in next post test cases would be written for each data replication method.
-Guru