Welcome to GraphEngine Discussions! #343

The biggest (and the only) difference between a GE server and GE proxy is that: a server owns a data partition, a proxy does not. A typical user scenario is that we use proxies as query aggregators. Here is an example,

In the distributed setting, the data is partitioned and stored on many machines and a query needs to be executed on many machines. In this case, the proxy can serve as a middle layer between a GE client and the GE servers: the proxy takes the query submitted by the client, splits it into sub-queries that can be executed on different machines, dispatches the sub-queries to the machines, then collects the results of the sub-queries, aggregates the results of sub-queries, and sends the final result to the client.

Bin Shao (@shaobin) Perfect! I think that makes the GE Proxy a fantastic candidate for a type of decentralized query agent.

For those interested, over the past 6 years, all my development projects have been and are based on GE. I have learned a lot through detailed code sifting, tracing, and debugging of how the GE is put together. It is awesome. One of the areas that I have found to be very productive is the GE Availability Group capability. I have tried other Microservice containers and tried to use GE in Docker, Kubernetes, and Service Fabric, and none of them work as well as the native capability built into GE. I don't think I have exhaustive knowledge and know-how in other Microservice container environments, so there is space for others to examine, research and experiment. I will write about and share my learning when I have time and space.