Blue/green deployment

Mindbody utilizes a cloud-based platform for the fitness and wellness industry, offering services such as client booking, scheduling, payments, marketing, and analytics. Their email marketing platform is built on an Aurora PostgreSQL cluster, currently at version 13.8, with a size of approximately 17 TB and a workload distribution of 80% reads and 20% writes. Mindbody faced scaling and performance challenges due to architectural limitations and increasing data demands, leading to all workloads being directed to the writer node. The average BufferCacheHitRatio was below 80%, indicating frequent disk access rather than cache hits, contributing to higher query latencies and I/O costs. To address these issues, Mindbody adopted Aurora Optimized Reads, which enhances caching capacity and improves latency and throughput for I/O-intensive workloads. Transitioning required upgrading the database cluster to version 14.9 or higher, and extensive testing was conducted in a proof-of-concept environment. The upgrade process involved a blue/green deployment strategy to minimize production disruption. After implementing Aurora Optimized Reads, Mindbody experienced significant performance improvements, including a 50% reduction in average daily CPU utilization and a 90% reduction in ReadIOPS. The AuroraOptimizedReadsCacheHitRatio indicated that 85% of read requests were served from the optimized cache. Cost analysis revealed a 23% reduction in monthly Aurora costs post-transition, with potential for further savings by downsizing instances.