Linux-based RDBMS for Occasionally Connected Replication

Question

I am looking for a database replication solution which will meet the following requirements:

• Work with Linux-compatible RDBMS (Atom 32-bit single core architecture)
• Work in an occasionally connected environment, with two-way data flow
• Handle adding peers on the fly without a database engine restart

Here are the details of my configuration: I am running multiple Linux based mini-servers at remote locations which often lose internet connection for up to a week at a time. These mini-servers record data from users, and should replicate that data to any peers they can see on the network (more than one may be deployed in a given vlan for redundancy). When connected to the internet, these mini-servers should synchronize with our central server, and data may flow from the mini-servers to the central server, or from the central server to the mini-servers.

I think SQL Server can handle most of this, but it doesn't run on Linux. Any ideas? I can do a fair amount of custom coding to get a solution working, but my current solution is my own custom replication system, which needless to say, doesn't hold up to enterprise software expectations...

Answer 1

I am going to take the lack of responses as an implicit 'no'. If anyone is interested, I am planning my own postgres spin off with dynamic multi-peer replication support within the next year.