Today it would be difficult to argue against a microservices architecture in the cloud, with Spring Boot of course. As we scaled we struggled with memory usage and the cost of running our platform became eyewatering. John will walk you through the options we looked at and the solutions we found.
We’ve always used Spring, after all, we worked closely with the founders for many years and know the community well. When I first saw Spring Boot I knew it was going to be the next thing so back in 2016 when I set up my new company Velo Payments I bought in my previous team and we started building out new bank-grade payments platform. Some 50 person-years of coding later and we’re in production with several banks making payments all over the world. The biggest single issue we’ve had with Spring has been memory usage and start-up times. When you’re starting dozens of services you need a lot of machine instances, multiply that by 3 or 4 again for test, QA, sand-box etc. and cost go up, seriously UP. In production, our only real cost is the infrastructure costs. With some work, we were able to cut the costs by around 90%. We looked at other solutions and did a lot of tuning, this talk will explain what we did.