Reducing compute costs by 25% using Amazon EC2 M7i-flex instances with Hertz

Hertz manages a fleet of hundreds of thousands of rental vehicles around the world. Franchisees who run local Hertz locations access the Hertz database to manage reservations through the Hertz rental application. The rental app includes full inventory management, reservation, and billing capabilities, making it a critical component of running Hertz operations locally.

After running it on premises for many years, Hertz migrated the rental application to AWS, initially using Amazon EC2 M5 instances. Then, when Hertz implemented a company-wide cost-optimization strategy, the company realized that the rental application was underusing compute resources during off-peak hours. Hertz was able to better optimize costs for the application by using M7i-flex instances, which are designed to seamlessly run the most common general-purpose workloads. “One of the critical components of our cost strategy is understanding how to optimize and sustain people, processes, and technology,” says Krish Sastry, director of cloud platform engineering at Hertz. “Our goal was to save on compute costs and generate more revenue by maintaining continuous uptime, and M7i-flex instances empowered us to reach that goal.”

On AWS, Hertz has the flexibility to choose instances with the best combination of performance and cost for its workload. Hertz chose a Flex instance type—M7i-flex instances—because these instances are designed for workloads that don’t fully use the CPU all the time but can accommodate bursts of demand. These instances work well for rental application, which has periods of low usage but can hit 6 billion requests per day during peak season. Because M7i-flex instances pass on the cost savings from using otherwise idle compute resources to customers, Hertz was able to lower costs while using latest-generation performance. “Given the application’s predictable usage patterns, Flex offered an easy path to optimization without requiring any rework,” says Sastry.

Hertz tested M7i-flex instances before migrating its production environment, planning the cutover so it would not impact the customer experience. “The actual migration was very straightforward,” says Anoop Upadhyay, principal cloud engineer at Hertz.

Since migrating to M7i-flex instances, Hertz has improved performance while using fewer computing resources. The rental application handles up to 6 billion requests daily with 24/7 uptime, now with reduced error rates, faster response times, and reduced operational intervention after the migration. It also no longer requires manual rightsizing reviews. “We make sure that our stack is highly available so that we deliver a good customer experience,” says Sastry. “We are saving money and getting better performance using M7i-flex instances.”

By saving 25% on compute costs, Hertz’s infrastructure team is contributing to the company’s broader cost optimization strategy. “We contribute to the company’s optimization by providing continuous uptime and making sure that our stack adds value,” says Sastry.

These cost savings contribute to further innovation at Hertz. “The savings we achieved on M7i-flex instances give us lots of room to test and innovate with new technology,” says Upadhyay. “We are using our compute cost savings to give freedom to our engineers to innovate.” The company is also evaluating additional workloads in its AWS environment to migrate to Flex instances, particularly nonproduction, development, and test tiers, where burstable workloads align well with the Flex model.

Additionally, the company has empowered its teams to evaluate and test their own workloads for potential upgrades to new instance types. Hertz provides self-service environments and technical whitepapers with instructions for managing instances. “We want to give our teams the tools to be successful on the cloud,” says Sastry. “We are successfully running this business-critical system on M7i-flex instances, and we will continue migrating and modernizing our solutions on AWS over the coming years.”