Picking the nits in EC2 fleet management that we either didn't have time for before the Deloitte workshop, or discovered during it.

Success condition:

• The EC2 fleet management automation has a pleasant administrative UI with no glaring easily-fixed problems;
• Everyone waiting to use the probcomp stack on the cloud has a machine; and
• Everyone using the probcomp stack on the cloud is migrated to cloudformation-managed instances.

Explicitly out of scope:

• Cost management;
• Auto-scaling;
• Cluster computing;
• Making it easy for someone to run the stack on their infrastructure or their AWS account; and
• Rearranging the fleet to work around weird AWS resource limits (instead, ask for those limits to be raised).