Do not disable last active node in pool

That's an interesting question. Are you simply worried that a monitor will fail but that the application still works? Strictly speaking, I believe the monitor is as important as the load balancing functionality itself. In fact, I often refer to monitors as "the forgotten API" because of the shear flexibility that they give you. A well-built monitor should reflect, as closely as possible, the systemic functionality of the application that it, well, monitors. So that if a monitor fails an application pool member, there's a darn good reason for it.

As far as what to do if all of the pool members fail (I mean actually fail), then there are, figuratively speaking, tons of options. Just off the top of my head, you can:

1. Redirect users to an alternate site

    

2. Send traffic to an alternate pool

    

3. Display a maintenance page

    

4. Send email alerts to an admin that the problem exists

    

5. Attempt to restart the web services or other system tasks remotely

    

6. Use a global server load balancing device like GTM to send users to an alternate datacenter