CosmosDB Graph gremlin tips

After fiddling with the requests for inserting data using Gremlin, I decided to check on some other common calls and I was not super impressed with the results.

Looks like both limit() and count() steps are translated in not so optimal SQL statements that actually bring the entire dataset (based on the filters you’ve specified) and then they get applied on the client side. So in a way,

The other step that behaves a bit unexpected is valueMap(). Similarly to above, this step is computed client side, where the entire document is returned and filtered according to the valueMap parameters locally, in the SDK

AS a workaround for these limitations, you can always use the SQL API to issue your query directly which should make thigs a lot faster.