top of page
  • Writer's pictureDaniel Rankov

Achieve 40% better performance and 20% lower cost with the AWS Graviton processor

Updated: Dec 7, 2022

48 of the top 50 Amazon Elastic Compute Cloud (Amazon EC2) customers use AWS Graviton processors for their workloads. In addition, thousands of customers, including Epic Games, DIRECTV, Intuit, Lyft, and Formula 1 are running production workloads on Graviton-based instances with significant performance gains and cost savings.

What is the AWS Graviton processor?

AWS designs AWS Graviton processors to deliver the best price-performance cloud workloads running in Amazon EC2.

Graviton2-based instances support a wide range of general-purpose, burstable, compute-optimized, memory-optimized, storage-optimized, and accelerated computing workloads, including application servers, microservices, high-performance computing (HPC), CPU-based machine learning (ML) inference, video encoding, electronic design automation, gaming, open-source databases, and in-memory caches.

Many AWS services, including Amazon Aurora, Amazon ElastiCache, Amazon EMR, AWS Lambda, and AWS Fargate, also support Graviton2-based instances for a fully managed experience with significant price performance benefits.

Which AWS managed services support the AWS Graviton processor?

(At the time of writing - Oct 2022)

A wide range of AWS managed services already support the AWS Graviton processor. Let’s review some of them:

  • Amazon Elastic Compute Cloud (Amazon EC2) - AWS provides more than 550 EC2 instance types. We can select different processors (AWS Graviton, AMD, Intel); GPU based; bare metal instances, and such with enhanced networking and storage.

  • AWS Lambda - Lambda functions based on Graviton can deliver up to 34% better price-performance

  • Amazon Aurora and Amazon Relational Database Service (RDS) - engines such as PostgreSQL, MySQL, and MariaDB can provide a fully managed relational database service using Graviton-based instances. Graviton2 instances provide up to 35% performance improvement and up to 52% price-performance improvement for RDS open-source databases based on AWS internal testing of workloads with varying characteristics of compute and memory requirements.

  • Amazon ElastiCache - up to a 45% price/performance improvement over previous generation instances

  • Amazon OpenSearch Service - support Graviton on multiple instance sizes - m6g, c6g, r6g, r6gd

How hard is it to migrate to Graviton?

For the AWS managed services like RDS, ElastiCache, and OpenSearch, it’s a click of a button. For example, RDS migration can be described as “I select my database, and I click Modify”. Simple as that!

Are there any additional guides?

Graviton Workshop will walk you through these practical steps:

  • EKS cluster with multi-architecture (x86-64 and arm64) nodegroups

  • ECS cluster with sample task and service running on Graviton2 instance type

  • CI pipeline for multi-architecture docker container images using CodePipeline, CodeBuild, CodeCommit, and ECR (with docker manifests)

  • CI pipeline for running .NET Core 5 on Amazon EKS cluster with Graviton2 worker nodes

  • RDS migration scenario from MySQL 8 on m5 instance type to MySQL on m6g instance type

  • RDS migration scenario from MySQL 5 on m5 instance type -> in-place major version upgrade MySQL 8 -> to in-place instance change to m6g instance type

  • EMR cluster with sample ETL Spark job running on Graviton2 instance type

  • Amazon OpenSearch Service cluster migration scenario from the m5 to the m6g instance families, with minimal downtime.

  • Testing the performance of an application on Graviton2-based instances and evaluating the impact of changes.

AWS Graviton Fast Start program helps you quickly and easily move your workloads to AWS Graviton in hours for many applications such as serverless, containerized, database, and caching.

Graviton3 instances are up to 60% less energy for the same performance as a comparable EC2 instance.

For further information on how to optimize your workload - contact us.



Commenting has been turned off.
bottom of page