TicketSauce hits 99.99% uptime, cuts IT cost 30% migrating Rackspace to AWS

Ticketing traffic is adversarial to fixed capacity: a platform can idle for days and then take its month’s peak load in the first minutes of a popular on-sale. On the legacy private-cloud host, capacity was whatever had been provisioned in advance — scaling meant tickets, lead time, and over-buying. Moving to AWS was the opportunity to re-platform the application itself: containerize it, give it a runtime that scales on demand signal rather than forecasts, and hold it to an uptime bar worthy of a payment-processing workload — 99.99%, measured through the spikes, not around them.

Inside the four-week migration window, VeUP containerized the application and made Amazon ECS on EC2 capacity providers the production runtime. ECS service Auto Scaling tracks demand and grows the service through on-sale bursts, with the capacity provider managing the underlying instance fleet; an Application Load Balancer distributes traffic and health checks tasks, with AWS Global Accelerator in front for stable, performant global entry. Session state moved to Amazon ElastiCache (Redis) so tasks stay stateless and disposable — the property that makes scale-out actually work — while the data tier runs on Amazon RDS Multi-AZ. The whole environment (VPC, security groups, multi-AZ layout, the ECS services themselves) is defined in Terraform, giving TicketSauce repeatable environments and provisioning measured in minutes. Amazon CloudWatch closes the loop with the metrics the platform is judged on: availability, latency, and scaling behavior through peak events.