During the 2018 re:Invent, AWS announced the release of Graviton, as a new and revolutionary option for cloud-hosted processing services. This processor, built with 64-bit ARM Neoverse cores, offers a high level of efficiency and performance, making it a highly recommended option for a wide range of services and applications.
One of the key features of the Graviton processor is its architecture, specifically designed to support the demands of workloads in the cloud. Its highly scalable design allows it to easily handle high levels of concurrency, supporting the bandwidth that modern applications need. It also has a highly energy-efficient design, which helps reduce costs and improve the sustainability of cloud computing.
Since then, AWS has released 2 new versions of the Graviton processors.
- In late 2019, AWS releases the second version of Graviton, Graviton 2
- In the mid-2022, AWS releases the third version of Graviton, Graviton 3
GRAVITON TECHNICAL VIEW
Each central processing unit has large level 1 and 2 caches, so part of its workload will be managed by the cache, avoiding having to access memory and thus improving response times.
There is a physical separation between each vCPU core, allowing for greater isolation, as resources are not shared between vCPUs, except for the last level cache and the memory system.
When applications share data, that is, when they need to move from core to core very quickly, the inter-core connection technology, which consists of a mesh with 2TB/s of bisection bandwidth, improves interaction speed.
WHEN TO USE GRAVITON
The Graviton processor is a valuable alternative for organisations as it is compatible with various operating systems (Amazon Linux 2, Red Hat Enterprise Linux, Ubuntu, CentOS, Debian, among others) and it natively integrates with AWS applications and services, among which are:
- Containers: Amazon EKS, Amazon ECS, Amazon ECR
- Monitoring: Amazon CloudWatch,
- Security: Amazon Inspector, AWS System Manager
- CI/CD tools: AWS CodeCommit, AWS CodeBuild, AWS CodePipeline
- Databases: MariaDB, MongoDB, Memcached, MySQL, PostgreSQL
Its processing capacity also makes it ideal for managing web and application server-based workloads, providing the performance and efficiency needed to support high levels of traffic. Other use cases, thanks to its ability to handle large amounts of data quickly, include machine learning, data analysis, and IoT.
WHERE IT CAN IMPROVE
Although an increase in the adoption rate of ARM architecture is appreciated in recent months, there are still some details that must be considered when incorporating these architectures into our development systems and projects.
For example, if we are working within a development team that mainly uses x86 architectures (Intel or AMD), the artefacts generated may not be directly compatible with ARM architectures, which forces us to pay special attention to the source and destination architecture in order to generate artefacts that are compatible and executable on the destination architecture (Of course, the incorporation of a CI/CD tool in our development pipeline can make this problem disappear and be transparent).
Another aspect to highlight, and similar to the previous one, is that we may need to incorporate third-party libraries in the developments that do not have a version prepared for ARM architectures, which will cause extra effort when developing for this type of processors.
On January 25, 2023, AWS has announced the release of Porting Advisor, an open source tool that analyzes source code and generates a report with libraries and/or code that should be ported, including recommendations, which facilitates the transition to AWS Graviton-based instances.
ITS MAIN BENEFIT: COST SAVINGS
The main benefit of using this processor is cost savings, without penalising performance. When we talk about cloud-hosted systems, most services are pay-per-use, so it is necessary to reduce consumption as much as possible. In terms of cost savings, this processor is an excellent choice because it is highly energy efficient so it can help reduce the cost of running workloads in the cloud, particularly for applications that require high levels of computational power. In addition, this processor costs less compared to other processors that are available on AWS.
In this table, some examples of the percentage savings by using r6g instances with Graviton2 processors can be observed, compared to the r6a and r6i instances with x86 processors:
The AWS Graviton processor is an innovative technology that has the potential to greatly improve the performance and efficiency of cloud computing. With its scalable architecture, wide range of use cases and savings potential, the Graviton processor is an important addition to the AWS ecosystem and a valuable tool for organisations looking to optimise their cloud computing costs, always bearing in mind that a change of processor architecture may entail the revision and adaptation of application development.
Image: Unsplash | @alexkixa
Leave A Comment