We have a huge enterprise system, with geographically seperated SQL servers.
We have some requirements related with data replication between these servers.
We have considered the Sync Framework option, but there is a requirement regarding multiple parallel data replication. Thus, Sync framework does not seem to be an option.
Also, we do not have a central database; requirements state that user shall configure the topology and related conflict resolution steps. There are three conflict resolution options; "source wins", "destionation wins" and "timestamp based".
Also, note that, database structure is a bit complicated and we have some deep hierarchies.
Lastly, the only way of communication between these servers is using port 80; using other ports is not an option.
I need to decide how to implement the replication structure, but I do not seem to have any options other than creating a replication structure manually and implementing all features.
Do you know any tools, or have any ideas how to handle replication easily without any central SQL server and with user configurable topologies?
Best Answer
Almost all of your requirements can be provided by merge-replication. There is some information about it in my StairWay series here: http://www.sqlservercentral.com/stairway/72401/
Those articles don't cover replication using port 80 but that is possible.
However, the topology is always a tree. You have one root and the data is distributed through there. A real "distributed agents" topology, where each note can synch with each other node is not possible with the build in tools.