Does TigerGraph support eventual consistency?

#1

Hi, I’m very interested in building applications based on TigerGraph. Documents (Transaction Processing and ACID Support :: TigerGraph DB) say that TigerGraph supports strong consistency. However, there are some applications like GNN inference on dynamic graphs that might only require eventual consistency. Using strong consistency might lead to large transaction processing overhead. So I’d like to ask does TigerGraph support eventual consistency? Thanks!

#2

Hi @jiesun233 ,

We currently support strong consistency model, like what the documentation said. Due to Strong Consistency and Eventual Consistency being a different models of consistency, I don’t believe you can switch from one model to another - that’ll require A LOT of Engineering efforts.

As for your concerns about the performance of a strong consistency model in TigerGraph, I’d say that TigerGraph can be used to run GNN at scale - we had an internal testing of GNN model in TigerGraph a while ago at a scale of millions of vertices and hundreds of millions of edges.

Best,
Supawish Limprasert (Jim)

#3

Thanks for your response~

1 Like