Announcing Feature Workflows in LaunchDarkly
As we reflect back on 2019, we can't help but feel proud of how far we have come and how much we have seen our customers grow with LaunchDarkly. In addition to now serving over 2 trillion flags per day across thousands of customers, we've also rolled out product updates such as as Experimentation and Data Export that have unlocked a new set of use cases for our users.
With this growth, we have also had the opportunity to learn more about our customers, their teams' processes, and the tools they use for building and shipping software to end-users. We learned that while feature flags are immensely powerful, the journey between flagging a feature to shipping a significant new release can often be pretty complex. Some notable themes that stood out were: centralized ticketing systems to track approvals, manually pushing a slider every few hours for incremental rollouts, and staring at an APM chart for a spike that will determine your next move.
With these learnings in mind, we are happy to announce Feature Workflows—a product update that enhances our core platform and extends the value of feature management across your organization.
Introducing Feature Workflows
With Feature Workflows, you can now elevate your use of feature flags by creating complex workflows within LaunchDarkly that better map to how your teams build, ship, and control software. You can also define custom release strategies that seamlessly integrate LaunchDarkly with your organization's existing tools and processes.
Feature workflows start with three powerful building blocks: Scheduling, Approvals, and Metric Checks.
Scheduling in LaunchDarkly introduces the ability for users to pre-define and schedule their flag and segment changes for future points in time. While this sounds deceptively simple, it unlocks the potential to create a progressive delivery strategy by scheduling the incremental steps for releasing your feature to 100% of your user base.
For example, you may want to schedule the feature to turn on for internal testing two days from today, then enable it for your ‘beta' customer segment four days later, and finally start an automated rollout from 0% to 100% over five days. Now, you can define this feature workflow ahead of time and be assured that it will take effect at the specified date and time.
Bonus: An interesting use of this functionality is the ability to automatically remove users from a flag. This allows you to manage trials easily by granting users access to a feature for a limited time and then automatically remove them from the flag's targeting when their trial is up—no developer intervention required.
While LaunchDarkly can be immensely powerful to help control your app in production in real-time, it's understandable to want a second opinion to validate your changes. To help with this, we're introducing the ability for users to request or require approvals before making flag changes.
For example, you may want to get your teammate to review and sign off on the changes you're about to make to a flag in production. Now, you can request a peer review from within LaunchDarkly.
In addition to requesting peer approvals to gain confidence, admins can also approvals for certain environments, in which case users need approvals in order to make flag changes.
We also wanted to support teams that may already have change management practices in place or use third-party tools to manage changes to their production environment for compliance purposes. Therefore, our approval workflows will integrate with tools such as ServiceNow and Jira to allow users to manage requests from within them.
Feature flags become much more powerful when they integrate with the rest of your ecosystem to become aware of the real-time state of your software. To help teams maintain their desired states across operational metrics and business metrics, we have added metric checks that trigger dynamic changes to your flags when pre-defined thresholds are crossed.
For example, you could configure a flag to automatically turn off a feature if LaunchDarkly learns that the API error rate has exceeded a healthy threshold in Datadog. This gives your team the ability to resolve the issue without having to worry about the experience for your end-users.
Similarly, for teams that experiment to learn which feature variation could lead to the highest conversions, metric checks can allow them to automatically have the winning variation of an experiment rolled out to all their customers. With this functionality, you can offload the responsibility of “if x, then do y” onto LaunchDarkly to ensure that your feature releases are safeguarded to optimize for the metrics you care about.
Feature Workflow Templates
While scheduling, approvals, and metric checks are incredibly useful stand-alone, they become even more powerful when they are combined together to automate and share your team's feature management practices. With LaunchDarkly, you can now create workflow templates that match your team's release strategies and can be applied across many flags.
By creating these workflow templates, you can have them handy to apply to new flags as they're created and be assured that these features will follow your desired release strategy, whether it's a rollout, approval, metric check, or a combination of all them.
In summary, Feature Workflows allow you to:
- Ensure consistent best practices across teams
- Automate and safeguard changes to features
- Create standardized release processes within an organization
Want early access? Let us know.
We're excited about empowering LaunchDarkly's users with this next phase of feature management capabilities. Our product team welcomes your feedback as we get ready for the perfect launch. You will see these updates roll out to your LaunchDarkly account over the coming months, but if you're interested in getting early access to any of these features, you can sign up here.