Use Case: Migration to AWS Graviton Processors
Oleksandr Abharian
Scaling eCommerce on AWS | AWS Retail Competency | Founder & CEO of IT-Magic - AWS Advanced Consulting Partner | Helping eCommerce Brands Scale, Cut Costs & Handle Peak Traffic on AWS
There are two leaders on the market of CPUs: Intel and AMD. However, there are companies that want to make a difference and use their own processors. For example, since 2010, Apple has been utilizing their A-series chips for Apple Watches, iPhones, and iPads and in 2020, they ended their partnership with Intel on Macs by switching to their M-series chips.
Amazon Web Services also belong to one of those companies. In 2018, they announced their custom-designed processors that would cater to specific workloads in their cloud infrastructure.?
Let’s review what these processors offer to businesses and how IT-Magic transitioned one of our biggest clients to AWS Graviton.
What is AWS Graviton?
Unlike traditional x86 processors, Graviton processors are built on the Arm architecture and now offer more memory and resources for a lower price.
AWS Graviton processors are designed to strike a balance between performance and cost-effectiveness. They offer better power efficiency, which leads to cost savings for certain workloads that require continuous operation and scalability.
Graviton processors are particularly well-suited for scale-out architectures that handle increased workloads. This makes them ideal for web servers, microservices, containerized applications, and other distributed workloads.
The second iteration of Graviton processors is called Graviton2. It features significant improvements over the previous generation, including more cores, larger cache sizes, and enhanced single-threaded performance.
The latest addition to the Graviton processor family is AWS Graviton3. These processors offer up to 25% enhanced compute performance, 2x higher floating-point performance, and 2x faster cryptographic workload performance compared to AWS Graviton2.
By offering Graviton-based instances, AWS aims to provide customers with more choices and flexibility in selecting the most suitable infrastructure for their specific workloads.
Why is Graviton a good alternative to traditional processors?
AWS Graviton presents itself as a compelling alternative to traditional processors due to several key factors:
As you see, businesses can benefit from cost savings, increased scalability, and access to cutting-edge technology, all while contributing to a more sustainable and energy-efficient cloud environment.?
Amazon Web Services showcases many companies that have already transitioned to Graviton and got their benefits. For example, Wix received 15% lower workload cost with Graviton, Snap reduced CPU utilization by 10%, and Discovery observed 7x better performance in their Subscriber Management and the Media Asset Management platforms compared to x86 processors.
Case study: how IT-Magic moved the client’s servers to Graviton
The client came to IT-Magic with a request to save as much money as possible as the business was large and IT expenses took a significant part of its budget.?
Our first step was to migrate the system to AWS where all configurations were set up correctly including system right scaling and right sizing.
领英推荐
The second step was about a more profound cost optimization where we worked with the three key aspects: spot instances, reserved instances, and saving plans.
The results were great. The client’s system became cost-efficient, stable, and reliable. However, they still wanted to optimize costs. And this is when we thought about changing the processor.?
Moving the client’s servers from Intel to AWS Graviton was not a trivial task though not very problematic either. At the beginning everything went smoothly. Databases, Redis, Memcached and other services moved to AWS Graviton without issues. We changed instance types one to one (e.g. c5.large -> c6g.large) and monitored performance and load. The performance was great and the load decreased in sight from 60% to 50%. Along with these changes, the price started to decrease too.
When it came to Kubernetes, some complications occurred. By default, Docker images are built for the Intel infrastructure. To make Docker containers work with AWS Graviton, we had to modify the process of building images so that it was compatible with both, Intel and AWS Graviton.?
The key modification command was:?
docker buildx build --platform linux/amd64,linux/arm64,linux/arm/v7
Read more about multi-arch images for Arm and x86 in the article from Docker
This way, we successfully transitioned the project to AWS Graviton2. This solution allowed the client to lower their AWS bill more and direct the money saved to business development.?
Overall, the client gained more benefits, namely:
The client was satisfied with the result:
If your business experiences increased workloads, AWS Graviton processors can become a good fit and an efficient solution for you. Contact me directly and we will discuss all AWS Graviton opportunities that are open to you.