Implementing Linkerd as a service mesh improved application resilience and reliability in a cloud hosting platform
- Cloud hosting platforms have transient network failures and require TLS encryption for every microservice
- Implementing application retry logic and MTLS through Linkerd improved reliability and resilience
- Linkerd as a service mesh provided a perfect place to develop internal SLOs, dashboarding, and alerting
- Buoyant Cloud subscription provided even better dashboard stats
The devops team was struggling to deliver reliable service due to transient network failures and the requirement for TLS encryption for every microservice. They considered moving to another cloud provider or redesigning the platform. However, implementing Linkerd as a service mesh solved their problems quickly and easily. The automatic proxy injection and MTLS implementation through Linkerd improved application resilience and reliability. Linkerd also provided a perfect place to develop internal SLOs, dashboarding, and alerting. The team even invested in a subscription to Buoyant Cloud for even better dashboard stats.
In this session, Mark Swarbrick, Bink's Head of Infrastructure, will share how Linkerd gave the fintech startup the confidence to partner up with Barclays and serve millions of customers in the UK. Around 2016, the Bink infrastructure team started containerizing their apps. Not long after they migrated to the cloud and moved their apps onto Kubernetes. Still in the early days, they experienced issues with the cloud's unstable networking infrastructure, leading to multiple issues including huge amounts of random TCP disconnects, UDP connections just going missing, and other faults. After a little research, they gave Linkerd a try and, lo and behold, as soon as they started experimenting with it, network faults caused by the instabilities dropped significantly. The timing could not have been better. The team was just starting conversations with Barclays and Linkerd's metrics allowed them to monitor their SLOs and agree to the bank's ambitious latency and success rated based SLAs.Click here to view captioning/translation in the MeetingPlay platform!