The Power of AWS Graviton Processors: A Deep Dive into Performance and Adoption

Introduction

Amazon Web Services (AWS) has introduced Graviton processors to provide better price-performance for their customers. In this deep technical blog, we will explore the performance of Graviton-based instances, how they compare to x86-based instances, and the various ways you can adopt these instances for your workloads.

Graviton Processors: A Performance Breakthrough

Graviton processors are custom-built by AWS to offer optimized performance for cloud workloads. They are based on the Arm architecture, which provides significant benefits in terms of power efficiency and cost-effectiveness. With Graviton2 and the recently launched Graviton3, AWS has raised the bar in terms of performance, making them suitable for a wide range of applications.

Graviton2-based instances offer up to 40% better price-performance compared to x86-based instances, while Graviton3-based instances further improve upon this, providing up to 60% better performance per vCPU compared to Graviton2. In certain workloads, such as encoding, Graviton3 instances have shown to be more than twice as fast as Graviton2 instances.

Adopting Graviton-based Instances

Graviton-based instances can be adopted for various workloads and are supported by a wide range of AWS managed services, container orchestrators, and the DevOps ecosystem. To ease adoption, consider starting with managed services, which require minimal effort in terms of application modifications. Some of the managed services that support Graviton-based instances include RDS, Elasticache, and Fargate.

Containerized Workloads

Graviton-based instances work seamlessly with containerized workloads. Major container orchestrators, registries, and container-optimized Linux distributions are all supported. However, to use Graviton-based instances for containers, you will need an Arm64 version of the container.

Serverless Workloads

AWS Lambda now supports Graviton-based instances, making it easier to migrate your serverless workloads. Most customers use high-level runtimes such as Java, Python, or Node.js, which don't require recompilation or rewriting. However, be aware of potential caveats with native modules in some languages.

Migrating Compiled Applications

Migrating compiled applications, such as those written in C or C++, to Graviton-based instances may seem daunting, but most code is high-level enough to not require significant modifications. However, if your application uses assembly or calls directly to intrinsics, more work may be needed.

Microsoft .NET and Windows Server

Microsoft .NET Core applications can be migrated to Graviton-based instances on Linux, while production support for Windows Server on Graviton is not yet available.

Graviton Supported Services

Conclusion

AWS Graviton processors offer an unparalleled price-performance advantage for cloud workloads. With the wide range of support available, it's never been easier to adopt Graviton-based instances for your applications. To learn more, refer to the AWS Graviton Technical Guide and start leveraging the power of Graviton processors for your workloads.

To learn more, visit https://aws.amazon.com/ec2/graviton

Best Practises to build and optimize on Graviton: https://github.com/aws/aws-graviton-getting-started

AWS Graviton Fast Start: https://aws.amazon.com/ec2/graviton/fast-start/

AWS re:invent 2022 session : https://www.youtube.com/watch?v=lZkO-KelLnk

AWS Well-Architected Labs: https://www.wellarchitectedlabs.com/sustainability/100_labs/100_migrate_services_to_graviton/100_migrate_rds_to_graviton/