I’m not quite sure if I understand what you’re planning to do. Why would you install anything on the hypervisor itself? That’s actually (officially) not possible with ESXi (the free version), and probably not what you want, anyway. Consider the hypervisor as a kind of “big” BIOS. It allows you to connect your VI Client (the Windows application) to it to create and maintain your VMs, monitor their performance etc.
But it’s not really supposed to do anything more than that.

You would probably want to create and setup the Gateway VM first, and connect it to vSwitch0 (connected to the physical public NIC) and vSwitch1 (virtual connection to all other VMs). And then you can put any number of VMs on that box, and only connect them to vSwitch1, hence making them accessible only via your Gateway VM.
You can connect vSwitch1 to a second physical NIC, which faces your local offices network, as well, if you want to. In that case, the Gateway VM would be the only connection for the other VMs _and_ your office network to “world”. Is that what you are planning to do?

Cheers
Carsten

I’m using ESXi 3.5 u4 on HP Proliant ML110 G5.And I need to setup a small experimental VM setup in the way similar to yours with the Gateway VM ( ie your VM1 ) acting as a reverse proxy.

In my test setup I want the Gateway VM to be accessible from hypervisor via dbclient ( dropbear’s lightweight ssh client ). To be more clear the Gateway VM has two NICs ( one public facing & one internal facing ). Now the I want the internal facing IP range to be visilble from hypervisor. Is that possible. If you think it can be done can you suggest me way to do it?

And it will be really kind of you if you can enumerate the steps for the experimental setup you mentioned, so that I can follow how to do it?