- - Development of a go-based "feature service" which allows features to be managed by developers and product managers.
- - Co-ordination of feature state across microservices, without relying on deploys.
- - Feature rollout strategies (percentage based, market based or custom attribute based).
- - Performance considerations of our approach, how to handle large volumes of request traffic in go services.
- - Graceful degradation, how to handle failure when core services (such as features) are unavailable.
- - What has worked for Vend with feature rollouts, reducing the impact to customers and support teams.
Nick Walker / Platform Engineer - Vend
Nick works on the Platform team at Vend, responsible for ensuring that the system architecture scales for their rapid growth in customers and developer team size.
Over the past 3 years he's been working everything from internal tooling for microservices, Kafka-based databuses and routing layers for datacenter migrations.
His passions include providing Vend developers with GIFs for all occasions, where he is summoned on Slack by his handle "@needagif".
Talk / Releasing features in a dynamic microservice environment
A number of organisations use "feature management" to release new features to subsets of customers outside of normal deploy processes. These might include "release new feature A to 20% of customers per day" as a simple example. Releasing features in this way allows for slow rampup for load testing, usability and risk management purposes. Vend's microservice and rapid deployment environment (30+ releases per day) makes this type of feature management somewhat challenging, with multiple services needing to coordinate feature releases/management at the same time.
This talk will discuss Vends approach to feature management:
Wednesday 10th August 2016