In my projects I use arango to calculate most of the business logic and project data sets. At about 100k of JSON lines of resulting code, I experience low performance of data delivery. Guess this is mostly caused by the serialization process. Which is absolutly ok!

All my servers are optimized to deliver high loads of data by using zero-copy mechanism of netty. Thats why I would like to reduce the cpu usage by skipping the serialization process.

Is it somehow possible to get the very raw velocy stream? I would like to post huge results directly to the client and decode it there.

It would be perfect if I could somehow get the stream into a direct buffer. Since the driver does not use netty it's probably not possible.

If this issue seems too off topic please just close it.