We determined that multiple days of DevOps consultancy would be the best way to move forward with Perkbox. The first few days were spent on site at Perkbox’s London offices, where we met with their development team and discussed the most effective ways we would integrate with them. We performed an analysis of what they were doing at the time, the infrastructure they were using, and what they were looking to achieve.
Built in resilience at every level, with resources spread between multiple availability zones and correctly set up for replication and failover.
- Autoscaling in line with demand, meaning Perkbox never pay for more resources than they need, but also have additional capacity available when they need it – and all automated.
- Application level monitoring using Prometheus, to provide detailed metrics about bottlenecks within the application.
Upon investigation, the DC/OS platform wouldn’t provide the features & flexibility Perkbox required. Because of this, we recommended replacing DC/OS with Kubernetes and committed to spending the remaining days building a highly available Kubernetes platform on AWS.
The platform is fully automated using Terraform orchestration, meaning that they can use the same platform on different regions, or different AWS accounts.
We worked with Perkbox to help them design and deploy their services into Kubernetes, alongside deploying and managing a number of important supporting components. This included Vault for secret management and Prometheus for monitoring.
We worked alongside Perkbox to develop the orchestration code for surrounding AWS services in-line with industry best practices. This included cloud services such as Aurora, ElastiCache, Route53, CloudFront, & API Gateway.