Hi, yes - that's one of the possible and feasible options that we utilize with some of our clients. That's not the only option (there exist other architectures, including using a 3-node clustered runtime environment with clustered metadata server), and combination/hybrid approaches with even higher number of servers.
But what you are suggesting is possible and will work.
There are always pros and cons in choosing deployment architecture (including RTDM runtime high-availability and load balancing architecture), and there is no single silver bullet - you can't tell whether one option or another is better without a deep dive into a particular client situation, requirements, technical and organizational dependencies, priorities, etc.
One specific implication of the architecture you suggested is that every diagram change will need to be deployed twice, manually (on environment 1 and environment 2). It's an overhead, but also sometimes a benefit.
Not sure about your question ("How to achieve Load balancing for the above 2 runtime servers") - you have responded to that yourself, just one line ago ("Active-Active Setup with HTTP Load Balancer / F5") - that's how (on a high level). Or are you asking about more details of how to leverage F5 for that purpose?
-------
Dmitriy Alergant, Tier One Analytics