GAE at the time was running as an experimental/beta product. There was no support, there was no SLA, and there was no deprecation policy. When a product is experimental or beta, eventually the time comes around when we need to look at next steps.
We concluded that the pricing was insanely low at the time and did not make for a viable long-term product. And when we talked to customers, they absolutely didn't want us to discontinue the product, but instead they wanted us to invest in it and provide support.
So that's what we decided to do. In working with customers, we determined that (after reasonable optimizations), the vast majority of apps would experience 2-5x increase in bills. Obviously we lost some customers because of this, but much fewer than you think, and mostly (but not always) because they were simply architected in ways that badly matched GAE, or had business models that assumed a cost-of-operations that was unrealistic (either on GAE or anywhere else), and that we had in effect been subsidizing.
GAE targets being the "best" option, including cost - but you need to include the entire cost, and factor in ease of use, SRE operations, durability, scalability, etc. We're happy to duke it out with any apples-to-apples comparison for total cost of ownership.
