I’ve written a fair bit about the use of a Gateway, and there is now a 3rd reason as Satya points out in his 10 reasons to use a Gateway post. 10 is a bit of stretch, but he points out an excellent fact about the gateway beyond the original two drivers for Gateway use:

1. Minimizing use of certificates in a remote domain outside the trust boundary.
2. To minimize points of communication from a perimeter network behind a firewall.
3. To optimize bandwidth utilization. This contradicts the old school of thought as documented in the OpsMgr Design Guide, but MS has shown it to be true.

As mentioned, the Gateway can be used for high availability scenarios, with agents reporting to a Gateway with the ability to fail to another, and Gateways configured to failover to multiple Management Servers.

For step-by-step instructions in Gateway deployment, including high-availability configuration, see out Gateway and PKI Scenarios Guide, or check out the 5 part series , which offers short bits of guidance on the Gateway role.