I’m a DevOps engineer at Anchorfree. We collect a lot of data from our servers and clients since we have one of the most popular VPN services. Recently we faced a challenge to apply GDPR cleaning on our data, while at the same time we had to scale up to 20 billion messages/day (2TB of gzip-compressed per day). These requirements led us to major architectural changes after many iterations of prototypes and benchmarks. So I’d like to share my experience in the mentioned scope, good and bad architectural decisions under requirements of high load and GDPR.