I should have been more clear on why I chose to go with the WARP client/tunnel in our Manage Engine/Desktop Central scenario - we are dropping our ancient Cisco VPN solution (no slight to Cisco, their product is amazingly stable) for a more flexible solution due to the company’s move to support WFH more broadly.
We still utilize classic, locally hosted file sharing in our data centers around the world, so while Cloudflare’s WAF is great for web applications, we still rely on classic VPN for access to internal resources that are not web connected. Not to mention, we also have mesh ipsec tunnels between our sites as well.
As for Manage Engine’s Desktop Central, it works amazingly well over internal networks/ipsec tunnels. In addition to the main DC server we have 6 DC distribution servers handling over 10k clients. On the internal network there are zero issues getting timely updates and new configurations thanks to the distribution servers. Once a machine leaves the network, this is where the problems start - especially if you have a lot of machines.
MangeEngine provides a “Secure Gateway Server” for Desktop Central, that already acts like a reverse proxy. It’s the best way to secure your main DC server without exposing all ports. It’s actually the correct answer here as what you want to put behind WAF - not the DC server itself. (this way you only need to expose a single port, not all). More information here: Link
However - and this is why I recommend the tunnel - the “Secure Gateway Server” for Desktop Central was designed for a small remote workforce. The implementation was never intended to cover a large use case as it completely ignores the use of Desktop Central distribution servers. When COVID hit, within 2 weeks, most of the 10,000 internal machines became remote workers and we had several issues with the Secure Gateway Server being overloaded. Not to mention, the geographical issue of clients in different countries getting updates from our main DC server in the UK. Our infrastructure was not originally designed to handle this, let alone redirecting thousands of clients through one endpoint.
For one, without a VPN connection the Desktop Central agent will not be able to talk to any of the servers (distribution or main) without a hole punched in our firewall - and a prolonged disconnect on the agent causes significant delays in configuration changes in the long run.
Since we need access to internal resources anyways (AD authentication, classic file sharing and other non web based apps), the WARP tunnel serves as a great tool to access all of our Desktop Central distribution servers and main server on top of being a great replacement for Cisco.