How we reduced AWS Cloud bills of BotGauge by 28.6%

About the company:

BotGauge AI is a next-generation AI-powered test automation platform founded in 2024. By leveraging advanced in-house Natural Language Processing (NLP) technology, BotGauge allows teams to effortlessly create software test cases using plain English, enabling faster testing cycles, higher software quality, and significant cost savings. Trusted by early adopters in the technology sector, the company empowers businesses to democratize automation and accelerate digital innovation with a lean, dynamic approach.

Services Used:

• CloudVictor Optimize – Boost Your Bottom Line

Challenge:

As BotGauge's infrastructure diversified and scaled across different AWS components to handle growing data volumes, its OpenSearch cluster became a critical component for log ingestion, analysis & debugging. However, the existing cluster setup consisted of costly and underutilized resources, such as dedicated master instances and ultrawarm nodes, which were not optimally configured for BotGauge’s current business requirements. Additionally, the use of on-demand instances for core OpenSearch data nodes significantly increased monthly expenses.Given the nature of BotGauge’s workload—heavy ingestion, low query rate, and no strict real-time indexing SLA - it was clear that the OpenSearch architecture was over-provisioned & mis-configured, leading to unnecessary operational costs.

BotGauge partnered with CloudVictor to optimize their OpenSearch infrastructure while maintaining compliance requirements and operational resiliency.

Analysis:

Before crafting the optimization plan, CloudVictor’s AWS specialist team performed a deep technical analysis:

1. Data Storage:

• Log data was being accumulated in OpenSearch indexes without any cleanup Index Statement Management policy.
• No periodic index backups.
• No purging of old index data.
• Enough EBS HDD provisioned on the cluster data nodes to hold the required amount log data based on the compliance requirement.

1. Dedicated Cluster Master Nodes:

The OpenSearch cluster was using three dedicated master instances with over-provisioned instance size. Cluster master nodes’ metrics indicated minimal cluster management overhead. 

Fig: 243 CPU Utilization of 3 dedicated cluster master nodes.

• Cluster Data NodesCluster Data Nodes’ metric analysis showed that there is a decent amount of free CPU & memory resources on each cluster data node. 

Fig: 157 CPU Utilization of 3 cluster core data nodes.

Fig: 112 JVM Memory Pressure of 3 cluster core data nodes.

UltraWarm Nodes:Two ultrawarm nodes (with 2.5 TB of storage) were deployed to hold older, transitioned index data.

• On-Demand Data Nodes:Three r6g.large.search data nodes were running on on-demand pricing, even though workload characteristics were stable and predictable, making them ideal candidates for Reserved Instances to optimize long-term costs.

Solution:

Based on the detailed analysis, Cloud Victor implemented a targeted optimization plan:

• Data Management & Storage:We implemented Cleanup Index Statement Management (ISM) policies to automate cleanup of log data older than a certain threshold (based on compliance requirements).ISM policies for daily S3-based backup of index data were also implemented.
• Decommissioning of UltraWarm Nodes:With the ISM cleanup policies implemented, the entire index data was consolidated into the existing provisioned EBS HDD in the cluster’s core data nodes. This made the ultrawarm instances empty and redundant. We decommissioned both the ultrawarm instances without impacting data availability or performance, thereby saving on the monthly bill.
• Removal of Dedicated Master Instances:With the existing scale of log ingestion (indexing) & analytics (search queries) workload, cluster data nodes had a sufficient amount of free CPU and memory resources to absorb the cluster management task (owned by the cluster master nodes) without risk. We configured daily index snapshots to S3, ensuring rapid restoration, minimizing any potential risk in case of failures or cluster instability.Then we decommissioned the dedicated master instances and observed the cluster metrics for a few weeks (through configrued CloudWatch alarms) ensuring stability.
• Reserved Instances Adoption:We had a planning discussion with BotGauge team about the evolution & future planned use of the OpenSearch cluster infrastructure. Based on that discussion & EBS limits, we migrated the On-demand OpenSearch data nodes to OpenSearch Reserved Instances, locking in approximately 31.14% additional annual savings compared to the prior billing model.
• Risk Mitigation Strategies Implemented:
  • Three Data nodes distributed across three AWS Availability Zones to ensure high availability and quorum even during any failures/cluster instability.
  • ISM policies for Daily index snapshot to S3 for disaster recovery & archival purpose.
  • Disk space & cluster stability monitoring alarms were deployed to proactively monitor storage utilization/cluster instability risks.

Final Outcome:

By implementing these optimization strategies, we were able to significantly reduce BotGauge’s OpenSearch cluster monthly costs by 69.45% ($741.7 monthly savings) without any impact on log retention compliance, cluster stability, or performance.

We strengthened OpenSearch cluster’s Disaster Recovery through an automated daily index snapshot in cost-efficient archival storage (S3) and quick-restore strategy, ensuring operational resilience.

Business Impact:

• With these cost savings measures implemented, BotGauge was able to unlock cash flows of approx. USD 30,000 over the course of next 3 years which they will be investing into growth activities like hiring and influencer network expansion.

Built a future-ready foundation, capable of handling log data growth without exponential increases in cost or complexity.

At Cloud Victor, we specialize in optimizing cloud environments to maximize performance while minimizing costs. With over 12 years of experience architecting and managing complex AWS infrastructures, our team of AWS and Amazon alumni has designed scalable, high-performance solutions including Big Data ETL platforms, ML-powered search systems, and 3000+ TPS webservices. As an AWS Partner, we bring this deep expertise and proven methodologies to every engagement.

Whether you need to optimize your AWS spending, strengthen your cloud architecture, or future-proof your data platforms, our team has the knowledge and experience to help you achieve operational excellence and sustainable growth.

If you're looking to unlock hidden savings or improve your AWS infrastructure efficiency, just like BotGauge, reach out to us at Cloud Victor.

"“CloudVictor team enabled us to effortlessly optimize our AWS costs and guided our Devops team with clear, actionable insights.”"

Pramin Pradeep

Co-founder & CEO, BotGauge