Using feature flags and observability for service migrations

Change is inevitable, and that’s a good thing, especially in relation to software development, where it means delivering new and innovative features that improve user experience and quality of life. Additionally, in the case of service migration, change can mean improved performance and lower costs. But creating change reliably is no easy task, particularly when it comes to evolving architectures that make up today’s modern cloud environments, which are deeply complex and unpredictable.

If your platform goes down, your business suffers and your reliability comes into question, potentially tarnishing your reputation. Therefore, when approaching major architectural changes, devops teams should always ask themselves: How much work is involved in making this change? Is it worth it? 

Enterprise technology companies are tasked with maintaining both speed and reliability, requiring high-performance engineering practices. To improve application quality and performance for customers, the platforms and services these companies deliver must never experience diminished performance. All software vendors must rise to the challenge of continuously optimizing or risk being left behind for other more performant services. 

Each year, major cloud service providers release dozens, if not hundreds, of product updates and improvements – putting the onus on engineering teams to decipher which configuration optimizes cloud and application performance. But if there is even one issue in migrating to the new architecture, the likelihood of disruption increases dramatically.

Given the high stakes of these service migrations, engineering teams must meticulously plan their moves. To add to the high stakes of these migrations, the annual cadence of cloud feature releases is cause for concern, with over 90% of IT pros and executives reporting they’re worried about the rate of innovation among the top cloud providers and their ability to keep pace with it.

To keep up, organizations have implemented innovative approaches to service migrations – with one devops practice, feature management, gaining significant traction. In facing similar challenges to continuously improve our platform and interfaces, software developers have turned to feature management to continuously ship and release code, while maintaining stringent controls that allow for real-time experimentation, canary releases, and immediate code rollbacks should a bug cause issues.

For years, we have utilized the feature management platform LaunchDarkly to experiment, manage, and optimize software delivery; enabling a faster pace of innovation without compromising application reliability. Serverless functions make service migrations a snap, since changing which version of a function is invoked is simply a configuration change.

Experimentation but with the guardrails of observability & feature flags

Utilizing feature management, enterprise technology companies will be equipped to bring the same capabilities to their cloud optimization initiatives. The functionality of feature flags enables capabilities that can increase the pace of experimentation and testing, and allow enterprise technology companies to scale cloud architecture at the flip of a switch.  

Through experimentation, teams can troubleshoot issues – such as non-optimized code – that could result in delayed execution times. With feature flags, these releases can be quickly rolled back to restore normal behavior for users. With this amount of precision and control, teams can limit the duration and exposure of experiments, mitigating detrimental impact and helping to inform more cautious rollouts. Teams can then conduct follow-up experiments to ensure reliability and performance, while also utilizing continuous profiling to help troubleshoot the issue in their code. 

The control, speed and scale of these tests are only possible with feature management and observability. With feature flags, teams gain greater control to direct traffic to test environments, analyze performance, and quickly restore the original environment with no disruptions or downtime. In high-stakes situations such as these, engineering teams require solutions that can take the nerves out of their work and provide them with the capabilities they need to support continuous improvement initiatives and optimize their infrastructure.

More confidence to innovate

Feature flags and observability are for organizations large and small, traditional and cloud-native. Today, doing things the old-fashioned way often means doing it the hard way and, ultimately, slows innovation. By embracing devops techniques across software development and cloud engineering teams, organizations can take risks with the confidence necessary to truly innovate.

Pushing platforms to new heights often takes a concerted effort that otherwise would be impossible without the assurances that feature flags and observability provide. In adopting feature management for cloud optimization and migration initiatives, teams can be both fast and reliable, while also enabling a culture of constant experimentation and innovation.

Embracing new technologies and techniques to quicken the pace at which organizations can experiment, test and deploy new code or architectures is proving to be invaluable across industries. It’s time that high-stakes processes, such as deploying code in production and optimizing cloud infrastructure, become faster and easier – not just for our engineers, but also for customers who deserve the utmost in performance and reliability.

Liz Fong-Jones is Principal Developer Advocate at Honeycomb.