They claim millisecond latencies, global distribution, etc. Considering that it is multi-model, what advantages does Dgraph have over CosmosDB as a native graph database? Does anyone have any performance benchmarks?
Don’t have performance benchmarks. Dgraph is open source, while CosmosDB is a proprietary database. Dgraph is built from the ground up to be a graph database, while CosmosDB is most likely designed for SQL, and then retrofitted with NoSQL, Graphs, or anything else that comes out where Microsoft can compete.
You can see my take on Amazon Neptune:
CosmosDB is a strong product - no doubt. Our anecdotal experience of the performance has been mixed.
Unequivocally, it is expensive - especially beyond the 10 gig data limit for the first tier - and you will mostly likely need to write some code to track your “RUs” - request units. For most load operations, we have to throttle up, then throttle back upon completion.
By expensive, I means hundreds of dollars per month for a basic POC. If that’s a non-issue, it’s a good solution. Cost, however, was what lead us to look at Dgraph.
1 Like
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.