Ticket exchange site scales with demand on AWS
TicketSwap is among Europe’s biggest ticket exchange sites, helping customers across Europe buy tickets for sold-out events or sell extra tickets they no longer need. Founded in the Netherlands in 2012, the company previously operated its own server but found that it was unable to scale rapidly with demand and suffered from availability issues.
To remedy this, TicketSwap migrated to Amazon Web Services (AWS), attracted by its instant scalability and Europe-based hosting systems. The migration began by putting the TicketSwap database on Amazon Relational Database Service, instantly resolving management and maintenance issues, before moving its website over to AWS.
TicketSwap experiences huge changes in demand according to season and the proximity of events. For this reason, scalability and automation are of crucial importance, while load balancing is vital in order to gain the best value from its AWS services.
There are fewer events over autumn and winter, during which time the firm uses 10 nodes on Amazon Elastic Kubernetes Service (Amazon EKS) and Amazon Elastic Compute Engine. However, during the spring and summer demand can be ten times higher, arrive very quickly and be extremely short-lived.
In the immediate aftermath of major ticket sales, fans will flock to TicketSwap to set alerts and look for returns. After one event sold out in 15 minutes, huge demand saw AWS enable TicketSwap to instantly scale up to 15 times its normal capacity, before returning to normal half an hour later.
Since it first began using AWS, the company’s relationship with the system has evolved and its needs changed. TicketSwap now accesses finer-grained control with Amazon EKS and Amazon EC2 Spot Instances, while Amazon CloudWatch generates insights to help improve system efficiency.
Meanwhile, AWS Auto Scaling means it no longer pays for unused capacity. This proved vital to the company during the COVID-19 pandemic last year when most large events were cancelled and the company’s activity was dominated by smaller events.
