Graph database system to safely store data at high volume and scale


Techniques are disclosed to safely and performantly store data in a distributed graph database. In various embodi- ments, a combination of a replication protocol for data redundancy with a chain-commit protocol is used to ensure a safe ordering of concurrent updates across servers. The resulting protocol allows a distributed graph database to simultaneously uphold ACID properties and a useful degree of scalability through concurrent processing of updates in the typical case, and serialization of updates only where data integrity would otherwise be at risk.