This is a guest post by Balaji Narayanan, VP of Engineering at Clari, in partnership with AWS.
Enterprise revenue operations are inherently complex. Territories, teams, products, and strategies shift constantly and vary across organizations. Managing this complexity requires a structured system to capture, interpret, and act on revenue data. Clari defined the data infrastructure for enterprise revenue management, setting the foundation for a broader category now known as Revenue Orchestration Platforms (ROPs). ROPs consolidate key capabilities—conversation intelligence, sales engagement, pipeline management, and forecasting—into a unified system. They apply AI to help B2B teams execute revenue processes, engage buyers, and improve internal productivity.
Clari replaces manual reporting and static views with a real-time system that tracks changes in revenue data with full context. The platform unifies signals across the revenue tech stack to drive more accurate forecasting and faster decision-making.
The platform’s real-time nature demands constant database read and write operations throughout the business day, making it highly I/O intensive. With their growing customer base, Clari’s database operations were projected to increase three fold per year. This challenge demanded a new approach to database management that could handle their explosive data growth while keeping costs under control.
Amazon Aurora I/O-Optimized emerged as the ideal solution for Clari’s challenges. In this post, we show you how Clari optimized their database performance and reduced costs by 50% by switching to Amazon Aurora I/O-Optimized.
Solution overview
Amazon Aurora is a relational database system built for the cloud that provides unparalleled high performance and availability at global scale for PostgreSQL, MySQL, and DSQL. It provides two storage configurations: Aurora Standard and Aurora I/O-Optimized. Aurora I/O-Optimized provides enhanced performance with simplified pricing based on instance and storage usage, without Aurora Standard’s separate I/O operation charges. It is tailored for I/O-intensive applications like Clari’s. This pricing model proved particularly beneficial for Clari’s write-heavy workloads and real-time processing requirements. Aurora I/O-Optimized not only offered increased throughput and reduced latency to support Clari’s demanding workloads but also provided the perfect combination of performance and cost efficiency. The predictable pricing structure made it easier for Clari to estimate their database spend upfront, a crucial factor for managing their growing multi-tenant SaaS application.
The following diagram illustrates the solution architecture of the Clari Revenue Orchestration Platform. The Clari Revenue Orchestration Platform is built on a modern architecture that combines real-time data processing, intelligent automation, and applied AI to deliver comprehensive revenue operations capabilities. The platform is structured into three core layers that work together to drive business outcomes.
At the foundation is the Data Platform layer, which provides automated CRM synchronization, comprehensive revenue data integration, and custom hierarchies management. This layer ensures secure API access and maintains time-series snapshots for historical analysis and forecasting.
The Revenue Insights layer sits in the middle, powered by sophisticated multi-signal AI processing capabilities and purpose-built AI models. This layer includes an extensible AI/ML data lake that enables advanced analytics and predictive modeling.
The Revenue Cadences layer tops the architecture, delivering AI-guided workflows, intelligent task automation through AI agents, and automated recommended actions, all within a governed framework. This orchestration drives four key solutions: Pipeline Management & Prospecting, Sales Engagement & Productivity, Customer Retention & Growth, and Forecasting & Revenue Insights.
Aurora serves as the foundational database technology layer across Clari’s distributed architecture, powering multiple critical data stores and services. Aurora clusters support key components including UserDB for core user management, Datamart for analytical processing, Bootstrap for system initialization, and Activity-Meta for metadata operations. This strategic implementation of Aurora enables Clari to maintain high availability and scalability while handling complex data operations across various service clusters including Core Analytics, CRM data services, and Data Science workflows. Clari’s approach to scalable and resilient data management, supporting consistent performance across customer-facing applications and internal processing workloads.
Clari’s Journey to I/O Optimized
Clari had been using Aurora database for some time, but they discovered their new consumption-based model generated 100x more data and associated I/O. This increase was not a good fit for Aurora Standard’s cost model and created an urgent need to find a cost-effective solution that also maintained performance objectives.
Recent metrics showed one of the service clusters averaged between 38,000 IOPS and 55,000 IOPs per node over 1 month. Given the high I/O intensity for both reads and writes, continuing with Aurora Standard would be prohibitively expensive. Aurora I/O-Optimized emerged as a promising solution after a thorough cost analysis.
Clari discovered that the break-even point where it made financial sense to switch from Aurora standard to Aurora I/O-Optimized was at 4,000 IOPS. At 40,000 IOPS on average per cluster, they saw savings of approximately 72%. When IOPS increased to 50,000, their savings grew to 77%. These figures clearly demonstrate the cost-effectiveness of Aurora I/O-Optimized for high-IOPS workloads.
Clari’s implementation strategy focused on two key requirements: zero downtime and minimal code changes. The process of enabling Aurora I/O-Optimized can be done through a simple API call or by changing the setting to storage_io_optimized through the AWS management console. This adjustment is an online operation, meaning it could be performed without any downtime to their database clusters. Clari then used Terraform to reflect this change in their infrastructure-as-code, ensuring consistent configuration across their Aurora clusters.
The entire process required just one engineer and a mere hour of Terraform updates, demonstrating the simplicity of adoption even for complex, large-scale deployments. This efficient implementation process underscores how Aurora I/O-Optimized can be rapidly deployed to address pressing database challenges in enterprise environments.
Benefits
The following chart shows that before switching to Aurora I/O-Optimized, storage I/O accounted for half of Clari’s database costs. After implementing Aurora I/O-Optimized, Clari was able to significantly reduce the I/O cost and reduce overall database spending by 50%.
The adoption of Aurora I/O-Optimized transformed Clari’s data processing capabilities between July 2023 and November 2023. With a 3x increase in processing speed and the ability to handle 10x more data volume, Clari removed previously existing barriers to innovation. Before Aurora I/O-Optimized, database costs consumed 50% of their infrastructure budget, limiting their ability to invest in innovation. Post-implementation, Clari launched three new AI-driven forecasting models and expanded their data analytics offerings, innovations that were previously cost-prohibitive. This transformation exemplifies how choosing the right database solution can have far-reaching effects on a company’s ability to innovate and compete in the fast-paced tech industry.
Conclusion
In this post, we showed how Clari achieved two key outcomes: remarkable performance improvements and meaningful cost optimization. The enhanced I/O capabilities of Aurora enabled Clari to process large volumes of data with unprecedented speed and cost-efficiency. This not only resulted in faster response times for users but also opened new possibilities for data analysis and insight generation. The success of this implementation demonstrates how the right database solution can be a game-changer for businesses dealing with high-volume, time-sensitive data processing.
As organizations across industries continue to generate and analyze ever-larger volumes of data, technologies such as Aurora I/O-Optimized will play an increasingly crucial role. Clari’s experience illustrates that such solutions aren’t just about handling more data—they’re about unlocking new potential for innovation and competitive advantage.
To learn more about how Amazon Aurora I/O-Optimized can help optimize your database performance, visit the Amazon Aurora Storage page or share your thoughts and questions in the comments below