Comparison
Change freezevsRelease train
Change freeze
nothing may ship for three weeks over the peak period, so a hundred changes go out together on the day it lifts.
A period during which deployments are prohibited, usually around a commercially critical window. It reduces the rate of changes and increases the size of every one that follows, which trades a small chance of many small failures for a large chance of one big one. It is most defensible when the release process is genuinely risky, which makes it a symptom worth reading rather than a practice worth defending.
Full entry →Release train
the release goes out on Thursday whatever is ready, and anything that missed it waits for the next one.
Shipping on a fixed schedule, with whatever has been merged and is ready, rather than when a particular scope is complete. It removes the negotiation about dates and makes cadence predictable, which is why it survives in mobile and in anything with an expensive release process. Its cost is batching — a fortnightly train makes every release a bigger, riskier change than a continuously-deploying team ever has.
Full entry →