Boosting Frontend Application Availability: My Journey from 50% to 99.9%

Originally shared on LinkedIn

In my career across multiple organizations, I’ve faced significant challenges in improving the availability of frontend applications. I’ve seen apps with availability below 90% and crash-free sessions as low as ~50%. Through diligent work and strategic improvements, I’ve successfully elevated these metrics to an impressive 99.9%.

Here are the challenges I encountered and the solutions that worked, so others can leverage these strategies to enhance their application’s availability and user experience.

Availability & UX Improvement Tips

1. Add Alert Rules

Define precise alert rules and ensure they are tagged to the right stakeholders to minimize Mean Time to Acknowledge (MTTA) and Mean Time to Resolve (MTTR).

2. Implement Multi-Stage Rollouts & Self Healing

Utilize multi-stage rollouts at 5%, 10%, 25%, and 100%. Perform thorough analysis at each stage and compare with previous versions to identify deviations and new issues early.

3. Define Different Levels of Alerting Rules

Create different alerting rules with thresholds based on metrics like Largest Contentful Paint (LCP), crash-free sessions, global error counts, high-volume issues, and more.

4. Mandate End-to-End Testing

Enforce end-to-end testing for every new feature development to ensure the master branch remains stable.

5. Enforce Unit Testing with Minimum Coverage

Mandate unit testing with at least 80% coverage for new code to prevent breakages from future unintended side effects.

6. Alerting and Monitoring on Infrastructure Health

Set up alerting for potential infrastructure issues such as min/max pod counts, memory utilization, and CPU utilization to proactively address performance bottlenecks.

7. Implement Feature Flags

Use feature flags to enable and disable features dynamically, allowing for safer rollouts and quick rollbacks in case of issues.

8. Foster a Culture of Continuous Improvement

Every incident gives us learnings to prevent further incidents and automate processes to self-heal the application. Make sure to act upon action items immediately, and foster a culture of continuous improvement.

Let’s take the challenge and give a better web experience to every user of our applications! 🚀