[design] isolated classloader vs project separation to support older Hadoop versions

["Elek, Marton" <el...@apache.org>]

The goal: support older Hadoop versions (and Spark, Hive, Hbase, etc...)

The current "isolated classloader" based approach (aka. legacy jar) -- 
which makes it possible to run Ozone together with older Hadoop -- 
doesn't work in secure environments.

We are thinking about replacing it with a different approach (with 
organizing the Hadoop dependent code parts and protobuf in a different way).


This document is not a 100% traditional design doc, but summarizes the 
current approach (problem definition) and a possible new approach:

https://hackmd.io/Io78mzXBR0aJrbTxQSfjdg?view

Please let me know if you have any questions/comments.

Thanks
Marton

---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-dev-unsubscribe@hadoop.apache.org
For additional commands, e-mail: ozone-dev-help@hadoop.apache.org