I'm really trying to get my hands on SQL CMD. I'm thinking of having each of my databases loaded with a utility database, which will have the hallengren stored procedures installed. And I want to output the print out to file share, so I would be using SQL CMD to connect to the different servers to perform maintenance and output the print message from the hallengren sp's to a network log folder.

I don't really need the engine on the management box.

And the reason I'm not using SQL Server Agent is to minimize downtime in case there is a network failure in the cloud provider we use, or the host or Blocks fail that are used my the database VM. I'm planning on using mirroring for our production databases, and having the mirror in a different data centers that the primary (in AWS this would be different availability zones with the same regions). If a separate box is performing maintenance, (which can be we can periodically take a snapshot of), nothing would stop us from performing a manual or forced service failover in an emergency. The extra step of reloading msdb in an emergency situation is averted. I could be wrong, but the less steps you have to do in a disaster scenario, the less likely you are of making a mistake. I do realize an argument can be made for scripting a reload of the system databases as part of the recovery process.

The reason I'm not using AlwaysOn Availability Groups, is simply because of the cost of running enterprise. I'm trying to get the business to stay on the ball with licensing, and it seems that it's easier to lease VM's in the cloud that include standard licensing as part of the cost of the cloud VM, easier to justify in our corporate culture anyway.

Thanks