Customer Profile

Greenworks Tools offers a safer, quieter, and battery-based alternative to gas powered lawn and landscaping equipment.

Customer Challenge

Greenworks Tools showed consistent growth quarter to quarter with their online marketplace, but their workload used an Elastic Beanstalk backed wordpress and custom marketplace application that relied on a RDS-MySQL implementation. The amount of web traffic mixed with newer application features reduced the efficacy of Greenworks Tools’s infrastructure.

Proposed Solution & Architecture

Triumph began by evaluating current CPU and memory requirements for both the Elastic Beanstalk compute and the RDS-Mysql implementation. The t-instances Greenworks Tools used worked well for spiking workloads, but consistent performance at a cost efficient point was more important. After several interviews with application developers and historical analysis of the Cloudwatch Logs for RDS and Elastic Beanstalk, Graviton2 instances were chosen for their effective CPU performance versus price. Early on during tests, it was determined that the instance re-sizing would need to be coupled with new auto-scaling groups.

Greenworks would transition to Graviton2 instances using AWS Elastic Beanstalk’s immutable deployment method. RDS would also be changed to Graviton2 by leveraging RDS-MySQL’s Point-In-Time-Recovery (PITR) to maintain data consistency and keep with promised RPO/RTO. Then, the Graviton2 instances were evaluated over the course of the next month for performance issues.

Outcomes of Project & Success Metrics

Greenworks Tools’s resizing with the help of Graviton2 reduced costs overall by reliably handling the workload without CPU and memory spikes. Because of this, Greenworks Tools saw a reduction in marketplace errors even under high sustained loads. Coupled with new auto-scaling groups, Greenworks Tools can move forward with the help of Graviton.

Lessons Learned

Graviton can meet both performance and price depending on workload.

Summary Of Customer Environment

Cloud environment is native cloud. The entire stack is running on Amazon Web Services. Stack contains Codepipeline, Elastic Beanstalk in Immutable Multi-AZ format, and RDS-MySQL in Multi-AZ with Read-Replica.