Nhse o34 release340

[martinsumner]

Setup for Riak 3.4.0 release

[martinsumner]

Discussed in https://github.com/orgs/OpenRiak/discussions/39.

Final performance validation done using single-host testing. Information from that testing:

• On OTP 24 relative performance (i.e. time to complete fixed volume test) between Riak 3.2.6 and 3.4.0 is within 1%. Performance does not vary by more or less than the margin of error if there is no OTP change.
• On OTP 26, relative performance of Riak 3.4.0 when compared to Riak 3.2.6 (running OTP 24) when running a balanced load:
  • 5% improvement when using the HTTP API, and a non-leveled backend.
  • 10% improvement when using the PB API, and the leveled backend.
  • 15% improvement in performance when using the HTTP API, and the leveled backend.

This is inline with expectations after CPU-efficiency enhancements made, that were based on profiling with OTP 26.