Feature flags are becoming table stakes among top-performing development teams. Many teams are starting their journey with feature flags by going open source or developing their own homegrown tool. As teams get familiar with these basic tools, they soon hit a wall that prevents them from tackling meaty problems at scale.
Basic tools alone are inadequate for managing the lifecycle of flags, understanding impact through experimentation, and communicating through integrations throughout the management stack. While we’d love for every team to migrate from those tools onto Split for their basic feature flagging, we understand that some tools are deeply embedded and hard to replace. There’s also the possibility (gasp!) you like that tool and want to keep it. That’s why we’ve made the bold decision to open up Split’s Feature Data Platform to 3rd party and homegrown feature flag data.
In the last year, I’ve analyzed over 40 open source feature flagging solutions. In each case, there was a lack of language support and feature coverage, inevitably creating data silos. Most open source solutions lack the basic collaboration capabilities and downstream connectivity required to scale. Put simply, these solutions do not provide the control, insight, and actions Matthew discussed in his article about Split’s product roadmap.
Recognizing this problem, we set out to make it possible for teams to get the benefits of Split’s Feature Data Platform no matter where their feature flags reside.
To do this, Split’s platform had to be reimagined so each component could be isolated yet remain fully integrated. For example, it had to be possible to use Split’s experimentation capabilities without using Split’s feature flags. The Feature Data Platform announced today achieves all of this.
How it Works
Split can now ingest upstream data from nearly any existing feature flagging system including open source, homegrown or other third party providers. We call this capability External Feature Flags.
External Feature Flags is built through a combination of native integrations, synchronization tools, log tailers and SDK integrations, bundled together to create the next industry standard for feature flag data.
We’ve already built connectors for three feature flag sources – Unleash, Split.rb and Microsoft Azure App Configuration Feature Management. Once configured, feature flag data, including audit logs and impressions, will begin to appear in Split.
As this data flows into the Feature Data Platform you will be able to apply Split’s control, insight, and actions as though the feature data originated from within Split.
You can see these flags in Rollout Boards to better understand and manage your flag lifecycle.
You can see feature flag evaluations as they happen in near real-time with live tail.
You can provide a layer of governance with audit logs, capturing all feature flag changes to be able to see who made targeting changes and when.
And most importantly, you can pair these feature flags with customer event data to get the benefits of monitoring and experimentation.
In addition, External Feature Flags, when coupled with our Integrations Hub, enables sending feature level context to many of the other applications in your stack. Feature level context refers to who saw what feature at what time, and who made changes to a feature’s targeting rules.
With data flowing through the Feature Data Platform and out to Datadog, for example, your engineers can now correlate feature flag changes to performance indicators. Another option, when flags change they can trigger an alert via Slack. Those are only a couple of examples among the 25 integrations (and growing) available today in our Integrations Hub.
When I co-founded Split in 2015, I wanted to change the nature of how product engineering was done. Feature flagging and experimentation was how we did it. Over time we and our partners have seen the emergence of feature flags and experimentation as a new best practice in product engineering. I couldn’t be more thrilled to announce our upcoming integration with Microsoft Azure App Configuration feature management. Azure App Configuration feature management supports targeting customers by percentage, and Split’s integration brings the ability to view audit logs, impression records, and send flag data to downstream destinations.
Balan Subramanian, Partner Director of Product, Microsoft Developer Division at Microsoft Corp. added, “Customers can now use the feature flagging capabilities of Azure App Configuration seamlessly with Split’s advanced measurement and analytics, to better understand the impact that every feature change has on their end-users’ experience. This integration enables developers to deliver features quickly with confidence.”
We look forward to bringing this to market in the coming year.
My team and I would love to get your input, understand what additional libraries you are using and of course have you join us in our early access program. Sign up at https://split.io/eap.