May 25, 2021

Arm-based cloud computing is the next big thing: Introducing Arm on Oracle Cloud Infrastructure

By: Clay Magouyrk | Executive Vice President, Oracle Cloud Infrastructure Engineering


The market is changing, and Arm processors are now ubiquitous on smartphones, edge devices, and increasingly used in PCs, laptops, and servers, with an average of 22 billion Arm-based chips shipped each of the last three years.  This adoption was originally driven by Arm’s lower cost and lower power consumption, but Arm is now able to compete head to head on performance as well. Soon, Arm servers will be the outright performance leaders, with additional benefits such as performance consistency and increased security.

This architectural transition has significant implications on how we develop and run applications.  In decades past, there was significant competition between server CPU Architectures, with SPARC, MIPS, POWER, and x86 all competing.  For the past decade or two, X86 has become completely dominant on the server, removing the need for most applications and libraries to work across multiple architectures.  To take advantage of the current and upcoming Arm advantages, as an industry we have to invest in the Arm development and operating environment, rediscovering our multi-platform skills of the past.

Oracle is here to enable this transition.  To make this possible, we’re addressing every aspect of the challenge – offering the best Arm-based server platform possible; providing easy access to all developers and open source projects to develop and test on Arm; working with partners to enable the Arm community; and contributing to the Arm ecosystem directly with investments in some of the most important technologies in the industry.

Oracle’s Arm-based Server Platform

AWS was the first cloud provider to offer Arm instances with Graviton, and more recently, Graviton2.  They’ve done an outstanding job pioneering Arm for cloud providers.  At Oracle, we’re taking a different approach.  Rather than designing our own Arm server CPU, we’ve been working with Ampere.  Ampere is a CPU design startup, led by an amazing engineering team.  Ampere has designed an industry leading Arm-based server chip with Ampere Altra, which offers predictable performance with the most cores per socket.  Our strategy is to work with Ampere to design the best server Arm chip, enable it in our cloud, and enable customers to put it in their own datacenters.

The time is now for developers to build the next generation of applications with Arm.

I’m excited to announce our new Oracle Cloud Infrastructure Ampere A1 Compute platform, based on the Ampere® Altra® processor:

  • Available in a bare metal shape for maximum performance, with 160 single-threaded cores at up to 3.0 Ghz, and 1 TB of RAM. Each core is single-threaded by design with its own 64 KB L1 I-cache, 64 KB L1 D-cache, and a huge 1 MB L2 D-cache, delivering predictable performance 100% of the time.
  • Available in a flexible VM shape with our exclusive flexible architecture for precisely provisioning from 1-80 cores and 1-64 GB of memory per core - right-sizing your Arm workloads
  • Fully integrated with GitHub, GitLab, and Jenkins to enable seamless development and deployment into Ampere A1-backed compute or containers. With Oracle, Arm-optimized applications can run in the public cloud or dedicated regions in the customer premises.
  • The OCI Ampere A1 is the first “penny core” cloud server in the cloud at only $0.01 per core hour and $0.0015 per GB of RAM per hour.
  • And you can start for free with our Free Tier, with additional free resources available for qualified developers, ISVs, and universities from our new Arm Accelerator program.

Great price-performance for all your workloads

OCI’s Ampere A1 platform is the price-performance leader in the cloud. Based on industry-standard SPECrate2017 benchmarks, Ampere A1 provides an estimated 2.5x better price performance of comparable Arm-based server instances and estimated 4x or higher price-performance than comparable x86-based Intel Skylake or AMD Naples instances.

SPEC, SPEC CPU, and the benchmark SPECrate are registered trademarks of Standard Performance Evaluation Corporation (www.spec.org)

The Ampere Altra processor offers consistent performance per core and near linear scaling. Applications will benefit from lack of contention within a core, dedicated resources for each core, and a consistent high frequency clock speed. Altra’s also enables applications to scale performance linearly as you use more threads as compared to x86 based instances. This means you get more performance out of each Ampere A1 core. When running an encoding workload for example, you can see that performance scales linearly as cores are added.

A line graph showing the encoding on 8vCPU shapes.

The Ampere A1 is a general purpose platform and performs well in many use cases. NGINX is a good example of high throughput workloads like web servers, microservices, and API gateways. In Arm's testing of NGINX Reverse Proxy requests per second, Ampere A1 had 32% better price performance than comparable AMD Milan-based compute and 69% better price performance than comparable Intel Ice Lake-based compute.

A bar graph comparing Nginx reverse proxy with no auth between Ampere Altra, AMD Milan, and Intel Ice Lake.

“NGINX Plus for Arm-based deployments just got a lot easier. Developers using OCI Ampere A1 Compute instances can now quickly deploy NGINX Plus on both Arm and X86 platforms,” said Stuart Shader, Senior Business Development Manager at NGINX. “With this capability and using the Always Free Tier on OCI, customers can gain access to a cloud environment to deploy NGINX Plus Arm-based applications for as long as they want.”

Video encoding is a good example of a very CPU intensive workload. In Arm's testing of the popular H.264 codec, Ampere A1 had 22% better price performance than comparable AMD Milan compute, and 98% better price performance than comparable Intel Ice Lake compute.

A bar graph comparing the encoding price-performance of Ampere Altra, AMD Milan, and Intel Ice Lake.

Similar CPU-intensive workloads include AI inferencing. OnSpecta, a deep learning software provider, found that Ampere A1 offered 2-9X price performance advantage in a series of AI inferencing benchmarks.

A graphic depicting three bar charts comparing vgg_19, Inception v3, and ssd_resnet_50 price-performance.

“OnSpecta significantly accelerates the performance of AI inference workloads on your OCI Instance when deployed on the Ampere A1 compute Platform. Our Deep Learning Inference Engine (DLS) optimizes the performance of trained neural networks resulting in up to 10x lower latency, 10x higher throughput and appreciable cost savings. OnSpecta’s customers can use DLS for all types of AI workloads including Object Detection, Video Processing, Medical Image Applications and Recommendation Engines. They will experience as much as a 3x cost performance gain over alternative solutions when using OCI Arm-based Ampere A1 Compute shapes.” Said Indra Mohan, CEO, OnSpecta

When you combine Ampere A1 with OCI’s overall price-performance leadership in containers, storage, databases, and networking, customers have the opportunity to completely transform their applications and businesses.

Enabling everyone to have access to Arm on the Server

Our goal is to offer enough Arm resources to develop your first application, test Arm at real world scale, and even build or port a production application. There are three ways to get free access to Arm instances in OCI. To start you can access the industry’s most generous free tier offering with 4 cores and 24 GB of RAM of Ampere A1 Compute as part of our Always Free tier. Just sign up here, and you’ll get this along with more than 20 great Always Free services like 200 GB of block storage, 2 Autonomous Databases, and more.

Sign-up also gives you access to our 30-day Free Trial, which gives you $300 in credits. That’s enough to run larger Ampere A1 instances with 16 cores, 96 GB of RAM, and 500 GB of high performance block storage for a whole month, or connect your Always Free Ampere A1 to a wider range of database, analytics, data science, observability, and integration services.

Finally, we’re introducing the new Arm Accelerator to scale your Arm development projects. Qualified open source developers, researchers, and industry partners can get up to $30,000 in credits valid for 12 months to help expand the Arm ecosystem or solve the next scientific problem faster.

Working together to expand the Arm Ecosystem

We've invested in a range of amazing partners and popular Oracle technologies to make it easier to build on Arm. First, we worked closely with Ampere for over a year to develop the platform, and ensure support for key technologies. It started with a shared platform design effort, which progressed to tuning work as we brought it to life. The last step was running the platform through our battery of compliance, reliability and stress tests. Once the hardware was ready, teams across Oracle worked with Ampere and other industry partners to help ensure the different parts of the technology stack was supported on A1. Arm has also been a critical partner working with Ampere and OCI to test performance across many workloads.

We've thoroughly tested Oracle Java, GraalVM, Oracle Linux, and our Kubernetes service (Container Engine for Kubernetes) to help ensure stability and performance. We've invested over 80 engineering years to optimize Oracle Linux for Arm and we've updated our Oracle Linux Cloud Developer image with everything you need to build Arm applications including language SDKs, container tools, database clients, developer tool chains, frameworks, and utilities. This includes testing of Java SE, GraalVM EE, Instant Client, Maven, Eclipse, Terraform, Buildah, and Ansible. Both Java SE and GraalVM EE are free on Oracle Cloud.

When it comes to CI/CD, we've partnered with key providers like GitLab, GitHub, and Jenkins to provide tested Arm solutions that can be deployed with a click of a button, and the Continuous Delivery Foundation to help ensure an open approach.

“CI/CD for Arm-based development just got a lot easier. Teams using Oracle Cloud can now easily deploy GitLab using the ‘Deploy to Oracle Cloud’ button, allowing developers to quickly build and deploy applications on both Arm and X86 platforms,” said Mayank Tahilramani, Director of Partnerships and Alliances at GitLab.

“GitHub Actions provides developers powerful, flexible, secure automation right in GitHub. We are excited to work with Oracle to bring choice to developers by ensuring GitHub Actions is supported for Oracle Cloud’s Arm-based Ampere A1 Compute platform.” - Joe Bourne, Director of Product Management, GitHub

“We’re happy to announce the availability of Jenkins automation software on the Ampere A1 Compute platform for building, deploying, and automating projects on Oracle Cloud Infrastructure”, said Olivier Vernin, Jenkins Infrastructure Officer. "Now developers can use Jenkins and start building next generation of applications easily on the latest Arm-based processors in Oracle Cloud.”

“Support for Jenkins on Oracle Cloud Infrastructure is a great example of Oracle’s commitment to open-source projects and the developer community. Oracle provides excellent technology solutions to help their customers succeed and we’re excited to see Oracle join the CDF,” said Tracy Miranda, Continuous Delivery Foundation Executive Director. “Jenkins and Oracle Cloud Infrastructure will be an extremely powerful combination for organizations needing continuous integration and software automation support in an Arm environment.”

To simplify deployment and testing in distributed Kubernetes environments or Android, we've partnered with Rancher and Genymotion.

“Our goal is to extend our Arm support even deeper with the new release of SUSE Rancher by adding support for provisioning multi-arch OKE clusters. When it comes to Arm, we simply want to say yes to most use cases— including Oracle Cloud Infrastructure, on premises or at the edge with Oracle Cloud Hybrid options.”  —Keith Basil, vice president of product and cloud native infrastructure at SUSE. 

“The Ampere A1 Compute platform on OCI with Genymotion gives developers the ability to build, test and deploy natively all on the same infrastructure. It provides better accuracy, performance, code coverage, and density of virtual devices on servers. Genymotion customers who are using our platform for sales, demonstrations, VMI, gaming and social media use cases could experience a performance increase, in some cases as much as 10x, using OCI Arm-based Ampere A1 Compute shapes.” Said Tim Danford, CEO, Genymobile.”

And for Datadog customers, we've extended our partnership to support Ampere A1.

“With the Datadog Agent for Ampere A1 on OCI, Oracle customers can use Datadog to determine the best shapes for their workloads, track migrations from other compute platforms, and get a unified view into the health and performance of their environment,” said Jimmy Caputo, Director of Product Management at Datadog.

Get started on Arm today

We're so excited to see what you build on Arm and OCI. It’s easy to get started. You can sign up for a free OCI account and spin up Arm resources by clicking here. If you’re interested in more specific workloads, here are some click-to-deploy application stacks and tutorials:

For more information about developing on Arm and the OCI Ampere A1 platform:

Executive Vice President, Oracle Cloud Infrastructure Engineering

Clay runs the engineering and software development team for Oracle Cloud Infrastructure, based in Seattle.  He has experience building cloud systems at scale at Amazon and elsewhere and was one of the first members of the Oracle Cloud Infrastructure team, working on Oracle’s Generation 2 Cloud since its inception.

More about Clay Magouyrk
This is a syndicated post, view the original post here