HDDS-11766. DataNode to cache blockDataTable and lastChunkInfoTable

[jojochuang]

What changes were proposed in this pull request?

HDDS-11766. DataNode to cache blockDataTable and lastChunkInfoTable

Please describe your PR in detail:

• DataNode updating checksum of an open file is becoming the bottleneck for HBase/hsync workloads.
• The main bottleneck is reading from rocksdb and deserialize it to the corresponding protobuf and further to the helper Java class.
• To address this overhead, add a Guava cache in front of rocksdb operation so the deserialization is avoid.
• Serialization cost when the rocksdb is updated, still exists and need to addressed seperately.
• That said, the ycsb data loading workload throughput is on par with the one where checksum is disabled, which means the bottleneck is probably on the client side again.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-11766

How was this patch tested?

Ran on a small cluster with HBase YCSB workloads.