JFrog’s Journey with AWS Graviton

Every business strives to optimize operational costs and efficiency. In the DevOps world, where cloud-scale operations are the norm, this becomes even more critical. At JFrog, while delivering a robust and highly scalable SaaS solution to our customers, we are equally focused on optimizing operational costs and maximizing infrastructure efficiency. Our recent transition to AWS Graviton processors for Amazon Elastic Kubernetes Service (Amazon EKS) and other managed services like Amazon RDS workloads reflects our commitment to optimizing performance and cost efficiency particularly with JFrog Artifactory. This strategic move represents a key milestone in our global deployment strategy, enhancing both cost effectiveness and sustainability.

In this article, I’ll walk you through our journey to migrate infrastructure workloads to AWS Graviton, share lessons learned, and discuss the benefits to JFrog customers.

The JFrog Platform and services

The JFrog Software Supply Chain Platform is the backbone of modern DevOps and DevSecOps workflows, empowering organizations to build, manage, and distribute software at scale with speed, security, and confidence. Our journey to the cloud began with a vision to deliver the full power of the JFrog Platform as a service – available anywhere, anytime, at enterprise scale. As we’ve talked about before in our own cloud migration story, this transformation enabled us to meet customers where they are, whether in the cloud, on-prem, or across hybrid environments. Today, the JFrog Platform runs on a globally distributed SaaS architecture that’s designed for high availability, performance, and scalability.

Testing the waters

Artifactory, our robust binary repository manager, handles both CPU- and I/O-intensive workloads. Available as a SaaS, self-hosted, hybrid, and multi-cloud deployment, it powers critical software supply chains. To enhance performance and reduce operational costs, our CTO office spearheaded a migration to AWS Graviton processors. This transition was carefully engineered to maintain high performance across our distributed systems. Here’s a detailed technical deep dive into how we successfully executed this migration.

Rigorous assessment and planning

Migrating to AWS Graviton was a strategic decision that required meticulous planning and rigorous testing. We began by analyzing download metrics and comparing architecture trends (intel64 vs. arm64). This thorough evaluation was essential to ensure our solutions would run seamlessly on Graviton’s Arm-based architecture. We also conducted a comprehensive review of our marketplace offerings to ensure compatibility and meet performance benchmarks. This diligent approach gave our DevOps team the confidence to move forward with the migration while maintaining high service standards.

Mastering migration with DevOps best practices

At JFrog, adopting DevOps best practices was key to executing a smooth and efficient migration. Our build process—spanning Java, Go, and NodeJS—progressed seamlessly, powered by Helm Charts, Cross Compile, and BuildX.

Here are some of the critical strategies that ensured success:

• Feature Flags: Allowed for a safe, incremental rollout by toggling features on and off in production, enabling real-time system monitoring.
• Node Affinity: Optimized workload placement on AWS Graviton nodes for better performance and resource management.
• Multi-Architecture Support: Enabled different architectures to coexist during the transition, reducing risks and ensuring continuity.

By leveraging these best practices, we achieved a controlled, stable rollout across our global infrastructure while maintaining reliability for our customers.

Financial and operational impact

From a financial standpoint, migrating to Graviton processors delivered substantial cost savings. We achieved a 20% reduction in compute costs, which directly lowered our Total Cost of Ownership (TCO) by 20%. While transitioning Reserved Instances (RIs) from x86 to Graviton posed challenges, the reduced node requirements and funds freed up from our Savings Plan commitment ultimately provided greater financial flexibility.

What’s in it for customers

Our migration to Graviton brings several key benefits to our customers:

• Improved Performance: With Graviton’s efficiency, CPU utilization dropped by 20%, and network latency decreased by 10%, ensuring a faster and more reliable experience.
• Enhanced Cost Efficiency: Lower operational costs enable us to offer more competitive pricing for both our SaaS offerings and self-hosted solutions.
• Sustainability: By reducing our carbon footprint by 60%, this migration reinforces our commitment to sustainability, making JFrog not just a powerful choice, but an environmentally responsible one.

Hear the JFrog team share more details on the move to AWS Graviton in the following video and this case study.

Lessons learned

Through this process of migrating to AWS Graviton, the team learned some valuable lessons along the way. For instance, a key learning was that we should have transitioned the cost to Reserved Instances (RIs) for AWS Graviton. Prior to using AWS Graviton, we had an existing commitment to Reserved Instances. However, when we migrated to AWS Graviton instances, we didn’t transfer the RI commitment to match the AWS Graviton commitment. If we were to do this again, the following steps would be the better approach:

1. Begin with Reserved Instances (RIs) on x86 servers.
2. Conduct a proof of concept (POC) using a small number of AWS Graviton servers.
3. Once the POC is successful, negotiate the transfer of the RI commitment to AWS Graviton.

Future steps

Looking ahead, JFrog remains dedicated to exploring further advancements in Amazon EKS and AWS Graviton processors. Our journey is part of a broader vision—the “Liquid Software” vision—striving for seamless dev-to-production flow and continuous software innovation.

In conclusion, JFrog’s technical transition to AWS Graviton processors highlights our unwavering commitment to optimizing infrastructure and performance. By embracing cutting-edge technology, we aim to deliver unparalleled value, efficiency, and sustainability to our customers worldwide. Stay tuned as we continue our explorative journey with AWS, innovating for a better, faster, and greener future.

Learn more about our collaboration with AWS, and take an interactive tour of the JFrog Platform on AWS to see it in action!