this stateful set is installed and is taking around 20 GB of ram in our cluster, what Is this used for ?
Hi, of course! 
I asked the dev team to take a look at this topic as soon as they can.
Hi, I believe the RAM requirement of prometheus is proportional to the number of nodes/pods in a cluster. Does your cluster have a large number of these resources? The built-in Lens metrics provider may not be ideal for all situations. It may be advantageous to customize a different provider, like from the prometheus-community Helm repo, and point Lens to it rather than the built-in metrics provider.
so what is this metric-provider used for ? and what actions does it take on its own ?
In short the metrics provider collects data (like CPU, memory, network traffic) in your cluster so that you can monitor usage/performance by system/node/pod/container etc. When Lens is pointed to the provider it graphically displays the information for you:
The heart of the metrics provider is Prometheus.
Alright, I have 2 clusters running in my environment, one does not have this deployment at all, and this stateful-set got automatically setup in the cluster I have mentioned up, I already have a prometheus server running in my cluster, how do I make sure cluster points to that prometheus server ?
From the cluster settings there is a page called “Lens Metrics” or “Builtin Metrics Provider” (the name changed recently). Here you can install or uninstall the built-in Lens metrics provider components you have asked about.
There is also “Metrics”, where you can point Lens to your existing prometheus setup. Lens supports a few common metrics setups, and tries to auto-detect what is installed. You can also manually select one of the setups and specify the prometheus service address, if Lens can’t find it.