How LaunchDarkly works

LaunchDarkly uses a real-time streaming connection which ensures we can deliver lightning-fast feature management services for enterprises of any size.

LaunchDarkly architecture

Our streaming technology provides a consistent connection to your SDKs allowing you to instantly turn features on or off at a significantly lower cost than polling. A global CDN ensures the lowest latency without remote requests.

Redundancy

If your application cannot contact LaunchDarkly, whether due to a misconfiguration, a Fastly outage, or even an AWS outage, you can configure your application to still perform exactly as your users expect by using the locally cached flag values. For another level of safety, every LaunchDarkly flag is also configured in your code with a default value that can serve as a fallback in the unlikely event that the SDK cannot initialize.

Scale

Flags and rules are stored in the LaunchDarkly core service, deployed across multiple AWS availability zones. Flags and rules are also cached and updated across Fastly’s worldwide CDN every time you make a change. Because flags and evaluation rules are cached locally, your application evaluates all requests without connecting to the LaunchDarkly service.

Consistency

LaunchDarkly uses server-sent events (SSE), a protocol for one-way real-time messaging, to send messages to your servers whenever you change feature flag rules on your dashboard. The SSE connection is handled automatically by our SDKs. No more waiting and wondering if the changes you requested have actually propagated across all your servers.

We are independently audited and SOC 2 Type II certified.