Posted December 14, 2023

Feature Experimentation: Performance updates to data build service for faster kill switches and better developer experiences

With our new datafile build service, customers will experience better performance and reliability when delivering feature flags and experiment changes.

Tom Burford
Tom Burford
Director, Product Marketing, Optimizely
decorative yellow lines on background

Optimizely Feature Experimentation users can now benefit from an average of 87% faster data file updates. The ability to generate data files in a faster and more predictable manner enables our customers to make updates to feature flags and experiments more quickly and reliably.

  1. Datafile build service – Performance, stability
  2. Webhooks by environment – Lower latency across all environments. Push notification that a new datafile is ready
  3. Secure environments - Security

Key features

  • Smoother workflow 
    It lets you update feature flags and experiments faster and more consistently as a seamless workflow step. 
  • Better developer experience 
    Developers can expect faster and more predictable feedback when configuring feature flags during local development.
  • Faster execution 
    Product teams benefit from “kill switches” to roll back problematic features and flawed experiments to protect user experience and conversion rates. 

Finally...

Speed, performance, and usability are key to delivering a better experience, and as such we are always striving to improve the performance of back-end services. Our improved datafile build service enables you to deliver feature flags and experiment changes to your end-users more quickly and reliably.

Optimizely Feature Experimentation generates a JSON datafile that represents the state of an environment in a customer’s Feature Experimentation project, this datafile is polled for and consumed by our SDKs to enable user-level decisions and tracking.

With our new datafile build service, Feature Experimentation customers will experience better performance and reliability when delivering feature flags and experiment changes to end-users. 

text

About the author