I think it's a terrible idea for the DoD (of all departments) to move all their infrastructure to any commercial cloud, Amazon or Oracle. So much critical infrastructure in the hands of a single company. What could possibly go wrong? They get hacked, experience significant outages, or worse, go out of business, and poof there goes all of the DoD's infrastructure... Whoever wins the bid will definitely become "too big to fail".
I see you aren't familiar with the DoD's (non-critical) infrastructure. Self hosted services go down for not insignificant periods of time not infrequently. Half the time if something goes down and the wrong person is out of the office, it doesn't get fixed until they come back.
I think fragmentation is the problem.
(Disclaimer: I've been out of the defense contracting business for some time)
It kind of shocks me that the DoD isn't looking for a vendor-agnostic implementation and then using the resources of multiple vendors. Not only is it key for resiliency from failure, but also it guarantees price competition, whereas being dependent on one vendor lets them jack up the price for all future transactions.
Vendor agnostic solutions come with overhead. Check out the PCF sizing tool. Running just the EC2 instances required to _manage_ a production environment starts at $30k/year per region using 1 year RIs. That doesn't include the Diego compute nodes that actually run your workload.
What do you get out of the deal? Software that makes AWS look like...AWS.
All three vendors appear to have settled on Kubernetes as the next level of abstraction. It would be great to see dollars that would have otherwise gone to Oracle enhance the capabilities of k8s.
