Prometheus metric on publisher message compression ratio

[jcass8695]

Is your feature request related to a problem? Please describe.
In our organisation we sink Pulsar messages for some topics directly to S3. It is useful for us to know the compression ratio that we are achieving on our payload size so that we are maximising efficiency of data storage, as we are able to grow/shrink the payload size to optimise for this.

Describe the solution you'd like
A prometheus histogram or gauge metric that producers publish on each message compression.

Describe alternatives you've considered
We considered calculating this metric ourselves, however the compression code is in the internal directory in the project. I do not think there are any public methods that would give us this data.

Additional context
N/A

[geniusjoe]

I think there is no direct compression ratio ralated prometheus metrics in go sdk.

Maybe we need to use these different compression algorithms to send same messages to different topics, and compare each topic pulsar_storage_size or pulsar_storage_logical_size prometheus metrics in pulsar broker.

I find some benchmark below and I think maybe there is no big compression ratio difference between these algorithms:

Compression Algorithm	Compression Ratio	Compression Speed	Decompression Speed
ZLIB 1.2.11 -1	2.743	110 MB/sec	400 MB/sec
LZ4 1.8.1	2.101	750 MB/sec	3,700 MB/sec
ZSTD 1.3.4-1	2.877	470 MB/sec	1,380 MB/sec
Snappy 1.1.4	2.091	530 MB/sec	1,800 MB/sec