The Partial Data Sandboxes delivered in Winter 2014 are a great start, but, for product development purposes, 5GB is more than we need — and much more than we can afford.
In a contemporary continuous development environment, the preferred approach is to have a feature sandbox for each feature under development, which corresponds to a feature branch in a Git repository.
Usng standard continuous integration systems with the Migration Toolkit, we have everything we need to create a modern continuous delivery environment — except the ability to easily clone a set of test data at an affordable cost.
It would be absolutely AWESOME if the Sandbox Template technology were applied to Developer sandboxes, so that we could copy even 100 records per object from an EE test org. (Rather than the 10,000 that is factored into the cost of a partial data sandbox.)
For development purposes, we don’t need more data — we need the ability to copy a small, well-engineered data set from a master org, quickly and easily, and then easily tweak the master data set as our product evolves.