• Does the push job require Kafka? By looking at the code and

this presentation▾

, I understood it is optional.

Venice does have a dependency Kafka. Kafka is used as the pipe through which all data, both batch and nearline, are funneled. The diagrams in the talk you linked are high level, trying to teach the basic concepts, but don’t go deeper than that into the implementation details… We have considered building a mechanism for the push job to write to Venice without going through Kafka, but that is just at the stage of thoughtware and there is no plan to build it for now…