International insurance company realized $200,000 in savings through cost optimization

Building a Minimum Viable Product for AWS cost control

The engagement with SoftwareOne began in late 2021 when SoftwareOne and the customer’s centralized IT team collaborated to build a Minimum Viable Product (MVP) for AWS cost control. The objective was to develop a tool that would provide the company’s top management with visibility into their AWS costs, accounts, product owners, and cost centers. This would enable them to make informed decisions and drive cost-saving initiatives across the organization.

SoftwareOne developed a comprehensive and user-friendly dashboard for the insurance company, providing an overview of their current costs over the past 30 days and potential savings going forward. The dashboard displayed key metrics, such as total costs, cost breakdown by account and product owner, and monthly savings potential. It allowed management to identify accounts and product owners consuming the most AWS resources, facilitating targeted cost-saving discussions.

The dashboard also offered drill-down capabilities, enabling management to analyze specific accounts and their consumption patterns in more detail. By anonymizing the account names, the tool provided insights into cost centers and product owners associated with the highest costs. This allowed for a focused dialogue with the respective account owners and product owners to understand their cost drivers and explore opportunities for optimization.

Upon reviewing the dashboard, SoftwareOne quickly identified areas for cost savings. SoftwareOne’s consultants engaged with the major cost accounts’ owners and initiated a dialogue to discuss their costs and explore potential savings opportunities. SoftwareOne’s team facilitated conversations with the account owners, product owners, and relevant teams to understand the value of their products and suggest alternative cost-saving approaches.

Identifying cost-saving opportunities

The initial cost optimization efforts resulted in savings of approximately $200,000 per year across the company’s various accounts. This encouraged further collaboration and discussions to explore additional cost-saving measures.

SoftwareOne’s team delved deeper into the company’s infrastructure and processes, seeking opportunities for fine-tuning and optimization. During the exploration phase, it discovered anomalies related to the insurance company’s Machine Learning (ML) entity. They realized that the entity was training models on 128-core virtual machines without proper cost-control measures. This led to skyrocketing compute costs, significantly impacting the company’s budget.

Cost Reduction through bundling, scheduling, and AWS spot instances

SoftwareOne worked closely with the ML entity to understand its requirements and processes and suggested bundling and scheduling the training of models to optimize costs. By leveraging AWS spot instances, which offer idle resources at significantly reduced prices, the company could reduce its compute costs for model training by 90%. This change in the process enabled them to train models without significant cost implications, resulting in more efficient and cost-effective operations.

The insurance company benefited from SoftwareOne’s cost management expertise and the implementation of the AWS cost control dashboard. It gained better visibility into its AWS costs, increased cost accountability among product owners, and improved communication with the business. The dialogue and collaboration between SoftwareOne, product owners, and management teams enhanced the company’s cost optimization initiatives.

Management appreciated the impact on the bottom line and valued the cost savings from SoftwareOne's guidance, which fostered a culture of continuous cost management improvement. Regular dialogues with product and account owners ensured cost-saving measures were integrated into processes, securing sustained, long-term savings.