Billions connected.
Global data distribution.
Global data replication.

Social network database

You run Facebook or Twitter or Whatsapp, Netflix, Snapchat or something similar on a global scale? Then you have always the task to manage your subscriber data in a way, that enables fast access on request from your users. Distributing a number of copies across the planet is a way to do so, but how to keep updates of a local copy consistent with the remote ones? Yes, automatic replication is the answer, a thing known long before we came up with our solution. But replicating everything is not feasible due to network bandwidth contraints and cost restrictions. So we implemented a lot more than just traditional replication.
The TeraCortex database solves this problem by flexible redundancy. This mechanism gives you a fine grain control over all aspects of replication:
  • Which copies to involve
  • Which objects to replicate
  • Synchronous or asynchronous
  • Compressed or not
  • Encrypted or not
  • Transactional or not
These properties are configurable per category of data. For instance one would choose asynchronous replication for fat images or videos or high speed updates. This hides the replication network latency from the updating client. One would replicate volatile data only at the local site, maybe uncompressed when that's faster. One would use transactional synchronous mode for updates, that must be consistent under all circumstances, like financial data.