Rollout Strategies
Percentage-based rollout
DeployRamp rolls out new code gradually. A flag typically starts at a low percentage (for example 5% of users) and ramps up automatically through a series of stages until it reaches 100%. This gives you confidence that new code works at scale before every user sees it.
Deterministic hashing
The SDK uses deterministic hashing to decide whether a user falls inside the rollout percentage. It computes hash(flagName:userId) and maps the result to a value between 0 and 100. Because the hash is deterministic, the same user always gets the same result for the same flag, so there is no flickering between the old and new paths across page loads or sessions.
Automatic error monitoring
DeployRamp monitors error rates in real time while a rollout is in progress. If the error rate spikes above the configured threshold, the rollout pauses automatically so the issue doesn't spread to more users.
If the threshold is breached, the flag is rolled back and flipped to the safe path, with no redeploy needed. Users are seamlessly routed back to the old code while your team investigates.
Manual override
You can set a flag's rollout to 0% or 100% from the dashboard at any time. This is useful for emergency kill-switches or for instantly enabling a feature that has already been validated in staging.
Staged rollout timeline
A typical staged rollout follows this progression:
- 5%: initial canary. Catch obvious regressions early.
- 25%: broader exposure. Validate under moderate load.
- 50%: half of traffic. Confirm metrics are healthy.
- 100%: full rollout. The new code is live for everyone.
Each stage is held for a configurable window (e.g. 1 hour, 1 day) before advancing. If errors are elevated during any stage, the rollout pauses and optionally rolls back.