Meet the Latest Amazon EC2 Family Members Featuring Intel Processors – M7i and M7i-Flex

Introduction

There’s a new Amazon Web Services (AWS) family in town, and it’s here to help boost performance and save you money. Compared to previous Intel processor-based instances such as M4, M5, and M6i, the new Amazon Elastic Cloud Compute (EC2) M7i and M7i-flex instances feature 4th Gen Intel® Xeon® Scalable processors to provide better performance per dollar for general-purpose workloads than previous generation instances. The M7i instances are the natural upgrade to the M6i, M5, and M4 instances featuring prior-generation Intel® processors, while the M7i-flex instances are a new, first-of-their-kind Amazon EC2 option bringing cost savings to the latest technology.

In this blog, the first of a three-part series, we’ll introduce these new instances, look at the new Intel Xeon processor in detail, and discuss the workloads best suited for these instances. In later parts of the series, we’ll dive into specific workloads and use results from internal testing to illustrate the kinds of performance improvements and cost savings you could see with your workloads.

The M7i Family

The M7i family, the latest generation of Amazon EC2 general-purpose instances, balances compute, memory, network, and disks to provide a solid ratio of resources for most workloads. Currently, the M7i family consists of nine instance sizes, from the m7i.large with 2 vCPUs to the m7i.48xlarge with 192 vCPUs. (In the future, the M7i family will also include bare-metal instances with 96 and 192 vCPUs.) 

Table 1 shows virtual M7i instances available now

Table 1: Details for M7i instances currently available.

The previous family, M6i, topped out at 128 vCPUs. If you have workloads running on the m6i.32xlarge instances that will soon need more resources, you can easily scale up to m7i.48 and avoid having to scale out by spinning up new instances. As of October 16, 2023, AWS On-Demand pricing in the US East region is $6.144 per hour for the m6i.32xlarge instances and $9.6768 per hour for the m7i.48xlarge instances, nearly the same rate in terms of dollars per vCPU. (1) The M7i instances offer several benefits over the M6i instances. They feature the latest in DDR5 memory for higher bandwidth than the previous generation, and they greatly increase the number of EBS volumes users can attach to each instance. Previous-generation instances allowed only 28 attachments, including network interfaces, EBS volumes, and NVMe volumes, meaning if you had two additional NICs attached, your EBS volume limit would drop to 26. New M7i instances allow up to 128 EBS volumes, regardless of the number of network interfaces or NVMe volumes also attached to the instance. While the exact number of EBS volumes your instance supports depends on the VM size, even the smallest VMs, with a 32 EBS volume limit, offer support for more volumes than any older instance. (2)

Flex Your Options with the Newest Amazon EC2 Offering

The M7i-flex family is an entirely new type of offering that is unique to the Intel M7i instance family that AWS has designed specifically to help customers save money. While some critical workloads demand that all resources be available all the time, other workloads do not. Previous generations of AWS instances required companies to overprovision and pay for all the instance’s resources, even ones they weren’t using. To avoid overpaying, customers had to right-size every workload and instance often paying for enough resources to handle peak demand even during off-peak hours. The new M7i-flex instances allow you to save money while ensuring your workload has resources for unexpected load increases.

AWS guarantees that M7i-flex instances will deliver full CPU performance 95% of the time, and at least 40% CPU performance in the remaining 5% of the time. (These percentages average usage across a 24-hour period.) (3) This allows AWS to offer customers a 5% discount for tolerating a possible dip in performance that comes from AWS re-allocating the capacity to other VMs. This makes these instances ideal for workloads that don’t need full compute capacity all of the time but have the potential for spikes in demand. According to AWS, M7i-flex instances provide up to 19% better price performance than previous M6i instances. (4)

 These flex instances range from 2 to 32 vCPUs, with network and storage bandwidth similar to their M7i counterparts. They also include the latest DDR5 technology for better memory bandwidth than earlier instances.

Table 2 shows the M7i-flex instance details.

Table 2: Details for M7i-flex instances currently available

Currently, the M7i and M7i-flex instances are available in a subset of regions, and AWS is expanding them to additional regions throughout 2023. Keep an eye on the AWS What’s New page for updates when new regions become available.(5) To give customers flexibility with budgeting, multiple AWS payment options—including On-Demand, Reserved, Spot, and Savings Plan—are available for both M7i and M7i-flex instances.(6)

4th Generation Intel Xeon Scalable Processors with Built-in Accelerators

Most knowledgeable cloud customers start their search for instances with the processor. 4th Gen Intel Xeon Scalable processors include several built-in accelerators to improve performance on critical workloads across AI, data analytics, networking, storage, and HPC. Because these accelerators optimize CPU core resources, they can boost power efficiency and help businesses achieve their sustainability goals.

All M7i and M7i-flex instances are available with latest-generation Intel Xeon processors with the Intel® Advanced Matrix Extensions (Intel® AMX) accelerator available to improve AI workload performance. Intel AMX enhances deep learning training and inference workloads when developers optimize their models to use the accelerator. Because Intel has enhanced open-source tools such as TensorFlow and PyTorch with their oneAPI Deep Neural Network Library (oneDNN), users can easily modify their AI workloads to use the Intel AMX instruction set.(7)  AWS has exposed the Intel AMX accelerator on its M7i and M7i-flex instances so that users can take advantage of this performance benefit. Intel testing on multiple deep learning models has shown that processors with Intel AMX can deliver performance up to 10 times that of earlier processors.(8)

 How this performance increase could impact your business depends on the type of AI models you use, but you could see faster response times with bot-driven customer chats, better predictive text for apps, and more. A future blog in this series will dive deeper into AI-specific workloads, so stay tuned.

The latest Intel Xeon processors also feature Intel Data Streaming Accelerator (DSA), Intel In-Memory Analytics Accelerator (IAA), and Intel Quick Assist Technology (QAT)—all of which can drive better performance across different workloads. Intel DSA helps improve CPU efficiency by offloading common data movement tasks to other components in the server. Intel IAA enhances in-memory tasks such as compression and encryption. And Intel QAT offloads tasks such as networking encryption and data compression to free cores and increase power efficiency. Support for these three accelerators is planned for bare-metal M7i instances coming soon.(9)

Workloads

Now that we’ve covered many of the benefits of M7i and M7i-flex instances, you may be wondering which workloads are best suited for them.

If you’re running a workload on an older Amazon EC2 instance generation such as M6i or M5, a quick lift and shift to an M7i instance could reveal immediate performance benefits and cost savings. Tests show that compared to M6i instances, new M7i instances deliver up to 40% higher transaction rates for MySQL OLTP databases(10) and as much as 43% better performance for CPU-intensive workloads.(11)

 If you’re moving a workload to the cloud, the best way to determine whether it is a good fit for M7i instances is to consider the resources it consumes. If the application is very memory-intensive but places relatively light demands on CPU or vice versa, another instance family might be more appropriate. However, most standard workloads—such as large applications and databases, gaming servers, and video streaming applications—require a balance of resources with high rates of CPU utilization and a good bit of RAM. Add the generous network and storage bandwidth of M7i instances and these instances are likely a good match for most of your public-facing and internal workloads.

Now let’s focus on M7i-flex instances, which we’ve explained are ideal for workloads that are not critical or have variable resource needs. What are some examples?

Many web applications and online services are multi-tiered applications that require separate services such as websites, databases, and so on. While the overall application might be resource-intensive, the web and application servers that host the URLs and web content often require fewer resources than the database. In this situation, you could reduce costs by hosting the less resource-intensive parts of your application on M7i-flex servers. This would allow you to pay less for the same capacity you typically provision to absorb spikes in demand. If a popular sale or viral news article were to drive a sudden increase in activity, the M7i-flex instances could handle it without application performance suffering. Microservices is another similar example; you can break larger applications into smaller components and place those that typically use fewer resources on M7i-flex instances.

Virtual desktop infrastructure is also well-suited for M7i-flex instances. Across the departments in your company, many users work primarily in web-based applications and a limited number of applications. If these users have enough memory on their desktop to handle these websites and applications, they don’t need excessive compute power. Using M7i-flex instances would give most of your users’ adequate resources at the 40% CPU baseline AWS guarantees. You would get the resources you need for power users and peak activity times while saving money.

Internal databases, applications, and batch-processing workloads can also benefit from less expensive instances that provide flexible resource usage. Even in very large companies, the number of employees using internal applications is relatively small compared to public-facing sites and applications. For many of these workloads, M7i-flex instances are a cost-effective way to ensure your employee applications have adequate resources. Preliminary testing shows that new M7i-flex instances offer 19% better price performance compared to M6i instances.(12) Later in this blog series, we’ll cover some of these workloads in more detail and present additional empirical data that shows the performance improvements you could see with M7i and M7i-flex instances.

Conclusion

M7i and M7i-flex, the newest Amazon EC2 instance families featuring 4th Generation Intel Xeon Scalable processors, introduce new performance and cost-saving opportunities. Take advantage of increased performance, new processor features, and better performance per dollar options by hosting your workloads on these new instances. And stay tuned for more in future blogs.

Watch this short video to learn more about the M7i instances.

(1) https://aws.amazon.com/ec2/pricing/on-demand/
(2) https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/volume_limits.html
(3) https://aws.amazon.com/blogs/aws/new-seventh-generation-general-purpose-amazon-ec2-instances-m7i-flex-and-m7i/
(4) https://www.intel.com/content/www/us/en/newsroom/news/4th-gen-intel-xeon-momentum-grows-in-cloud.html#gs.4hpul6
(5) https://aws.amazon.com/about-aws/whats-new/2023/08/amazon-ec2-m7i-flex-m7i-instances-aws-regions-europe/
(6) https://aws.amazon.com/blogs/aws/new-seventh-generation-general-purpose-amazon-ec2-instances-m7i-flex-and-m7i/
(7) https://www.intel.com/content/www/us/en/products/docs/accelerator-engines/advanced-matrix-extensions/ai-solution-brief.html
(8) https://www.intel.com/content/www/us/en/products/docs/accelerator-engines/advanced-matrix-extensions/ai-solution-brief.html
(9) https://aws.amazon.com/blogs/aws/new-seventh-generation-general-purpose-amazon-ec2-instances-m7i-flex-and-m7i/
(10) https://www.intel.com/content/www/us/en/content-details/794265/aws-ec2-m7i-instances-support-more-customers-and-more-transactions.html
(11) https://www.intel.com/content/www/us/en/content-details/794423/get-more-out-of-your-cpu-intensive-workloads-with-aws-ec2-m7i-instances.html
(12) https://aws.amazon.com/ec2/instance-types/m7i/

Notices and Disclaimers

Performance varies by use, configuration, and other factors. Learn more on the Performance Index site.
Performance results are based on testing as of dates shown in configurations and may not reflect all publicly available ​updates. See backup for configuration details. No product or component can be absolutely secure.
Your costs and results may vary.
Intel technologies may require enabled hardware, software, or service activation.
© Intel Corporation. Intel, the Intel logo, and other Intel marks are trademarks of Intel Corporation or its subsidiaries. Other names and brands may be claimed as the property of others.