Haste
2018 – 2021
75% lower infrastructure cost
Cutting infrastructure spend by three quarters
A latency-sensitive network product ran on hundreds of Linux servers across dozens of hosting providers around the world. The job was keeping that fleet consistent, observable, and affordable.
- Managed the fleet as code with Ansible playbooks for provisioning and routine maintenance
- Connected providers through a custom overlay network with bastions, SSH tunnels, and IPsec/OpenVPN
- Set up monitoring and paging with Datadog, Site24x7, Observium, and PagerDuty
- Built a client-server load-testing suite on AWS ECS so capacity decisions came from data
- Ansible
- AWS ECS
- Datadog
- PagerDuty
- Jenkins
- Linux