A definitive guide to releasing your best software

Software performance problems are costly from three vantage points. 

1. They require a great deal of time and money to resolve. When an incident occurs, several highly-paid engineers will scramble to find the source of the issue, write a fix, and rush that through their deployment pipeline. Or they’ll perform a rollback. Either way, you can begin to see why outages cost upwards of $9,000 a minute on average.
2. Outages divert focus from strategic initiatives—that is, they incur an opportunity cost. If your developers are constantly repairing broken services, they’re not building revenue-driving products. 
3. Outages lead to customer churn and bad customer ratings, making it harder for you to acquire customers. Depending on the severity and frequency, software incidents can deal a serious blow to your business.

Automatically remediate software issues

LaunchDarkly enables you to automatically uncover and remediate software performance problems before they impact customers—in real time. 

As a starting point, LaunchDarkly gives you a “kill switch” mechanism to ensure app stability. This lets you disable error-causing features in runtime. You don’t need to deploy code changes or roll your entire userbase back to an old version of your application. You toggle a feature flag (we’ll touch on flags later), and the problem is fixed. 

With Flag Triggers in LaunchDarkly, you can also turn ‘off’ error-causing features automatically. LaunchDarkly integrates with application performance monitoring (APM) and observability tools like Datadog, Dynatrace, Honeycomb, and others. When your APM detects a critical service issue, it triggers LaunchDarkly to disable the feature associated with the problem.

Upon this kill switch foundation, LaunchDarkly adds an even more robust, automated capability called Release Guardian—the enemy of outages, if you will. Release Guardian:

• De-risks releases through monitoring at the feature level
  Ensure that critical metrics are monitored on every release with connected events via LaunchDarkly’s API/SDKs and Sentry, an APM and error tracking tool. 
• Automatically detects and remediates errors
  Proactively alert on and disable software changes that negatively affect key operational metrics.
• Minimizes the blast radius
  Identify specific feature configurations that are causing errors for a subset of users (e.g., in the context of a progressive rollout). And perform an automatic change reversion before the feature degrades overall performance. 

LaunchDarkly thus enables you to maintain exceptional app stability, the second tenet of great software.

Customer success stories: App stability

• AMC Theatres avoids downtime during a high-volume event set off by the “Barbenheimer” craze. Learn about AMC’s success.
• Hewlett-Packard (HP) cuts down on rollbacks and reduces its mean time to recover (MTTR) from bugs. Learn about HP’s success.
• Atlassian improves its MTTR by 90% and improves customer satisfaction ratings. Learn about Atlassian’s success.

--

[2] “Calculating the Cost of Downtime,” Atlassian, accessed January 26, 2023.