This work was performed by engineers now on our team before Big Bad Byte was founded. One targeted RDS and Elasticsearch optimization produced $1.03M in annual savings. Across the full engagement, the telecom provider has saved more than $2M on AWS without compromising performance.
Challenge
The provider's AWS costs were increasing across production and non-production environments. The infrastructure included nine RDS databases and an over-provisioned Elasticsearch cluster. Cost reductions had to preserve system reliability and performance.
- RDS resources allocated beyond actual usage patterns
- Elasticsearch infrastructure sized above what was needed
- No room to trade production stability for a smaller bill
Solution
We analyzed usage patterns and cost reports, then changed the infrastructure in measured steps. The work covered storage pricing, instance scheduling and Elasticsearch capacity, with performance checked before capacity was removed.
RDS optimization
- Analyzed database usage patterns and cost reports
- Moved storage from io1 to gp3 while maintaining the same IOPS
- Scaled instances up for peak hours and down during off-peak periods
Annual savings: $560,000
Elasticsearch optimization
- Evaluated instance types and actual utilization
- Replaced higher-cost instances with more suitable alternatives
- Reduced data node count by 33% after in-depth testing
Annual savings: $470,000
Outcome
The RDS and Elasticsearch changes delivered $1,030,000 in annual savings. Further improvements across the engagement brought the provider's total AWS savings to more than $2M.
The result shows why cost optimization works best as an engineering process: measure the baseline, change one component at a time and verify that performance still holds.