Scaling SHIELD for more persistent SSH connections

[jhunt]

@thomasmitchell raised this concern in a discussion we had offline; moving it here so that we can discuss.

The active fabric presents a move to persistent agent connections - what is the strategy for scaling the SHIELD to handle more persistent connections?

[jhunt]

Let's set some goals on SHIELD scale size, for example: SHIELD should be able to handle 1000 connected agents, across 100 tenants, performing one hourly backup each.

Then, we can spin up infrastructure in our development lab (probably via a large-ish Kubernetes deployment) to place that demand / load on the core and see what happens. Upside: we can publish those findings as a blog on https://shieldproject.io!

[jhunt]

Our targets will be:

1000 connected agents across 100 tenants, performing one hourly backup each.