Show aggregate table metrics per cluster

[hopugop]

Describe the feature and the current behavior/state.
Table metrics are amazing. However, it only allows to choose from a single table at a time. It would be really useful to allow a view for all tables in a single view - think of it like a historical view of tables' operations, latency and dataset.

I recently used raw queries to achieve this, such as sum(scylla_column_family_total_disk_space{}) by(dc,ks,table) to aggregate disk utilization by table. There might be many others beneficial uses.

Who will benefit with this feature?
Anyone that is interested in historical metrics comparison for multiple tables.

[mykaul]

It makes sense to aggregate (sum? avg?) specific data from multiple tables - such as disk space. What else? Latency or ops make little sense to me.