I'm evaluating the use of RedisGraph vs Microsoft SQL Server 2019 graph features to build a social graph. One of the key criteria is high availability and cross data center replication of data. While Neo4j is a much-matured solution in general, I'm not accounting for it due to the cost & introduction of a new stack in the ecosystem. Also, due to the nature of data & org policies, using a cloud-based solution won't be a viable choice.
We currently use Redis as a cache & SQL server 2019 as RDBMS.
I'm inclining towards SQL server 2019 graph as we already use SQL server extensively as relational database & have all the HA and cross data center replication setup readily available. I still need to evaluate if it fulfills our need as a graph DB though, I also learned that SQL server 2019 is still a new player in the market and attempts to fit a graph-like query on top of a relational model (with node and edge tables). RedisGraph seems very promising. However, I'm not totally sure about HA, Graph data backup, cross-data center support.
When I looked at the graphing database solutions a few years ago, Neo4j won hands down. Plus, as a small company, I had access to it for free. Ask them about pricing, and it might actually end up being lowest cost. They now have a managed version called Neo4j Aura that I am very interested in so I don't have to manage the server. I haven't started using it. But a cursory investigation showed that Neo4j Aura was still probably the best alternative for my needs.