Proposal for Pre-Allocating and Caching Blocks in OM

[Penumudy Tanvi <ta...@cloudera.com.INVALID>]

Hi Ozone devs,

Frequent RPC calls to SCM for block allocation during writes can introduce
latency overhead and cause resource contention in OM, potentially becoming
a performance bottleneck under heavy workloads.

We have been working on the proposal outlined in HDDS-11894
<https://issues.apache.org/jira/browse/HDDS-11894> to prefetch blocks
asynchronously from SCM and cache them within OM to reduce this dependency
and enhance write performance, particularly during high spikes.

Your thoughts, concerns, or suggestions regarding this idea would be
immensely valuable.

Thanks & Best Regards,
Tanvi