New General Purpose, Compute Optimized, and Memory-Optimized Amazon EC2 Instances with Higher Packet-Processing Performance

{"value":"Today I would like to tell you about the next generation of Intel-powered general purpose, compute-optimized, and memory-optimized instances. All three of these instance families are powered by 3rd generation Intel Xeon Scalable processors (Ice Lake) running at 3.5 GHz, and are designed to support your data-intensive workloads with up to 200 Gbps of network bandwidth, the highest EBS performance in EC2 (up to 80 Gbps of bandwidth and up to 350,000 IOPS), and the ability to handle up to twice as many packets per second (PPS) as earlier instances.\n\n### ++New General Purpose (M6in/M6idn) Instances++\nThe original general purpose EC2 instance (**m1.small**) was [launched in 2006](https://aws.amazon.com/blogs/aws/amazon_ec2_beta/) and was the one and only instance type for a little over a year, until we launched the **m1.large** and **m1.xlarge** in late [2007](https://aws.amazon.com/articles/feature-guide-new-instance-types/). After that, we added the **m3** in [2012](https://aws.amazon.com/blogs/aws/new-ec2-second-generation-standard-instances-and-price-reductions-1/), **m4** in [2015](https://aws.amazon.com/blogs/aws/the-new-m4-instance-type-bonus-price-reduction-on-m3-c4/), and the first in a very long line of **m5** instances starting in [2017](https://aws.amazon.com/blogs/aws/m5-the-next-generation-of-general-purpose-ec2-instances/). The family tree branched in 2018 with the addition of the **m5d** instances with local NVMe storage.\n\nAnd that brings us to today, and to the new **m6in** and **m6idn** instances, both available in 9 sizes:\n\n\n\nThe **m6in** and **m6idn** instances are available in the US East (Ohio, N. Virginia) and Europe (Ireland) regions in On-Demand and Spot form. [Savings Plans](https://aws.amazon.com/cn/savingsplans/?trk=cndc-detail) and Reserved Instances are available.\n\n### ++New C6in Instances++\nBack in 2008 we [launched](https://aws.amazon.com/articles/feature-guide-amazon-ec2-high-cpu-instance-types/) the first in what would prove to be a very long line of [Amazon Elastic Compute Cloud (Amazon EC2)](https://aws.amazon.com/ec2/) instances designed to give you high compute performance and a higher ratio of CPU power to memory than the general purpose instances. Starting with those initial **c1** instances, we went on to launch cluster computing instances in [2010](https://aws.amazon.com/blogs/aws/the-new-amazon-ec2-instance-type-the-cluster-compute-instance/) (**cc1**) and [2011](https://aws.amazon.com/blogs/aws/next-generation-cluster-computing-on-amazon-ec2-the-cc2-instance-type/) (**cc2**), and then (once we got our naming figured out), multiple generations of compute-optimized instances powered by Intel processors: **c3** ([2013](https://aws.amazon.com/blogs/aws/a-generation-of-ec2-instances-for-compute-intensive-workloads/)), **c4** ([2015](https://aws.amazon.com/blogs/aws/ec2-instance-history/)), and **c5** ([2016](https://aws.amazon.com/blogs/aws/ec2-instance-type-update-t2-r4-f1-elastic-gpus-i3-c5/)). As our customers put these instances to use in environments where networking performance was starting to become a limiting factor, we introduced **c5n** instances with 100 Gbps networking in [2018](https://aws.amazon.com/blogs/aws/new-c5n-instances-with-100-gbps-networking/). We also broadened the c5 instance lineup by adding additional sizes (including bare metal), and instances with blazing-fast local NVMe storage.\n\nToday I am happy to announce the latest in our lineup of Intel-powered compute-optimized instances, the **c6in**, available in 9 sizes:\n\n\n\nThe **r6in** and **r6idn** instances are available in the US East (Ohio, N. Virginia), US West (Oregon), and Europe (Ireland) regions in On-Demand and Spot form. [Savings Plans](https://aws.amazon.com/cn/savingsplans/?trk=cndc-detail) and Reserved Instances are available.\n\n### ++Inside the Instances++\nAs you can probably guess from these specs and from the blog post that I wrote to launch the **c6in** instances, all of these new instance types have a lot in common. I’ll do a rare cut-and-paste from that post in order to reiterate all of the other cool features that are available to you:\n\n**Ice Lake Processors** – The 3rd generation Intel Xeon Scalable processors run at 3.5 GHz, and ([according to Intel](https://edc.intel.com/content/www/us/en/products/performance/benchmarks/3rd-generation-intel-xeon-scalable-processors/)) offer a 1.46x average performance gain over the prior generation. All-core [Intel Turbo Boost](https://www.intel.com/content/www/us/en/support/articles/000007359/processors/intel-core-processors.html) mode is enabled on all instance sizes up to and including the **12xlarg**e. On the larger sizes, you can control the C-states. Intel [Total Memory Encryption](https://www.intel.com/content/www/us/en/architecture-and-technology/total-memory-encryption-security-paper.html) (TME) is enabled, protecting instance memory with a single, transient 128-bit key generated at boot time within the processor.\n\n**NUMA** – Short for Non-Uniform Memory Access, this important architectural feature gives you the power to optimize for workloads where the majority of requests for a particular block of memory come from one of the processors, and that block is “closer” (architecturally speaking) to one of the processors. You can control processor affinity (and take advantage of NUMA) on the **24xlarge** and **32xlarge** instances.\n\n\n**Networking** – [Elastic Network Adapter](https://aws.amazon.com/about-aws/whats-new/2016/06/introducing-elastic-network-adapter-ena-the-next-generation-network-interface-for-ec2-instances/) (ENA) is available on all sizes of **m6in, m6idn, c6in, r6in**, and **r6idn** instances, and [Elastic Fabric Adapter](https://aws.amazon.com/hpc/efa/) (EFA) is available on the **32xlarge** instances. In order to make use of these adapters, you will need to make sure that your AMI includes the latest NVMe and ENA drivers. You can also make use of [Cluster Placement Groups](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/placement-groups.html).\n\n**io2 Block Express** – You can use all types of EBS volumes with these instances, including the io2 Block Express volumes that we launched earlier this year. As Channy shared in his post ([Amazon EBS io2 Block Express Volumes with Amazon EC2 R5b Instances Are Now Generally Available](https://aws.amazon.com/blogs/aws/amazon-ebs-io2-block-express-volumes-with-amazon-ec2-r5b-instances-are-now-generally-available/)), these volumes can be as large as 64 TiB, and can deliver up to 256,000 IOPS. As you can see from the tables above, you can use a **24xlarge** or **32xlarge** instance to achieve this level of performance.\n\n### ++Choosing the Right Instance++\nPrior to today’s launch, you could choose a **c5n, m5n**, or **r5n** instance to get the highest network bandwidth on an EC2 instance, or an **r5b** instance to have access to the highest EBS IOPS performance and high EBS bandwidth. Now, customers who need high networking or EBS performance can choose from a full portfolio of instances with different memory to vCPU ratio and instance storage options available, by selecting one of **c6in, m6in, m6idn, r6in**, or **r6idn** instances.\n\nThe higher performance of the** c6in** instances will allow you to scale your network intensive workloads that need a low memory to vCPU, such as network virtual appliances, caching servers, and gaming hosts.\n\nThe higher performance of **m6in** instances will allow you to scale your network and/or EBS intensive workloads such as data analytics, and telco applications including 5G User Plane Functions (UPF). You have the option to use the **m6idn** instance for workloads that benefit from low-latency local storage, such as high-performance file systems, or distributed web-scale in-memory caches.\n\nSimilarly, the higher network and EBS performance of the **r6in** instances will allow you to scale your network-intensive SQL, NoSQL, and in-memory database workloads, with the option to use the **r6idn** when you need low-latency local storage.\n\n— [Jeff](https://twitter.com/jeffbarr);\n\n\n\n### **[Jeff Barr](https://aws.amazon.com/blogs/aws/author/jbarr/)**\nJeff Barr is Chief Evangelist for Amazon Web Services. He started this blog in 2004 and has been writing posts just about non-stop ever since.\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n","render":"<p>Today I would like to tell you about the next generation of Intel-powered general purpose, compute-optimized, and memory-optimized instances. All three of these instance families are powered by 3rd generation Intel Xeon Scalable processors (Ice Lake) running at 3.5 GHz, and are designed to support your data-intensive workloads with up to 200 Gbps of network bandwidth, the highest EBS performance in EC2 (up to 80 Gbps of bandwidth and up to 350,000 IOPS), and the ability to handle up to twice as many packets per second (PPS) as earlier instances.</p>\n<h3><a id=\\"New_General_Purpose_M6inM6idn_Instances_2\\"></a><ins>New General Purpose (M6in/M6idn) Instances</ins></h3>\\n<p>The original general purpose EC2 instance (<strong>m1.small</strong>) was <a href=\\"https://aws.amazon.com/blogs/aws/amazon_ec2_beta/\\" target=\\"_blank\\">launched in 2006</a> and was the one and only instance type for a little over a year, until we launched the <strong>m1.large</strong> and <strong>m1.xlarge</strong> in late <a href=\\"https://aws.amazon.com/articles/feature-guide-new-instance-types/\\" target=\\"_blank\\">2007</a>. After that, we added the <strong>m3</strong> in <a href=\\"https://aws.amazon.com/blogs/aws/new-ec2-second-generation-standard-instances-and-price-reductions-1/\\" target=\\"_blank\\">2012</a>, <strong>m4</strong> in <a href=\\"https://aws.amazon.com/blogs/aws/the-new-m4-instance-type-bonus-price-reduction-on-m3-c4/\\" target=\\"_blank\\">2015</a>, and the first in a very long line of <strong>m5</strong> instances starting in <a href=\\"https://aws.amazon.com/blogs/aws/m5-the-next-generation-of-general-purpose-ec2-instances/\\" target=\\"_blank\\">2017</a>. The family tree branched in 2018 with the addition of the <strong>m5d</strong> instances with local NVMe storage.</p>\\n<p>And that brings us to today, and to the new <strong>m6in</strong> and <strong>m6idn</strong> instances, both available in 9 sizes:</p>\\n<p><img src=\\"https://dev-media.amazoncloud.cn/923bafc5d0a0473c8ebd089c0d493902_image.png\\" alt=\\"image.png\\" /></p>\n<p>The <strong>m6in</strong> and <strong>m6idn</strong> instances are available in the US East (Ohio, N. Virginia) and Europe (Ireland) regions in On-Demand and Spot form. [Savings Plans](https://aws.amazon.com/cn/savingsplans/?trk=cndc-detail) and Reserved Instances are available.</p>\\n<h3><a id=\\"New_C6in_Instances_11\\"></a><ins>New C6in Instances</ins></h3>\\n<p>Back in 2008 we <a href=\\"https://aws.amazon.com/articles/feature-guide-amazon-ec2-high-cpu-instance-types/\\" target=\\"_blank\\">launched</a> the first in what would prove to be a very long line of <a href=\\"https://aws.amazon.com/ec2/\\" target=\\"_blank\\">Amazon Elastic Compute Cloud (Amazon EC2)</a> instances designed to give you high compute performance and a higher ratio of CPU power to memory than the general purpose instances. Starting with those initial <strong>c1</strong> instances, we went on to launch cluster computing instances in <a href=\\"https://aws.amazon.com/blogs/aws/the-new-amazon-ec2-instance-type-the-cluster-compute-instance/\\" target=\\"_blank\\">2010</a> (<strong>cc1</strong>) and <a href=\\"https://aws.amazon.com/blogs/aws/next-generation-cluster-computing-on-amazon-ec2-the-cc2-instance-type/\\" target=\\"_blank\\">2011</a> (<strong>cc2</strong>), and then (once we got our naming figured out), multiple generations of compute-optimized instances powered by Intel processors: <strong>c3</strong> (<a href=\\"https://aws.amazon.com/blogs/aws/a-generation-of-ec2-instances-for-compute-intensive-workloads/\\" target=\\"_blank\\">2013</a>), <strong>c4</strong> (<a href=\\"https://aws.amazon.com/blogs/aws/ec2-instance-history/\\" target=\\"_blank\\">2015</a>), and <strong>c5</strong> (<a href=\\"https://aws.amazon.com/blogs/aws/ec2-instance-type-update-t2-r4-f1-elastic-gpus-i3-c5/\\" target=\\"_blank\\">2016</a>). As our customers put these instances to use in environments where networking performance was starting to become a limiting factor, we introduced <strong>c5n</strong> instances with 100 Gbps networking in <a href=\\"https://aws.amazon.com/blogs/aws/new-c5n-instances-with-100-gbps-networking/\\" target=\\"_blank\\">2018</a>. We also broadened the c5 instance lineup by adding additional sizes (including bare metal), and instances with blazing-fast local NVMe storage.</p>\\n<p>Today I am happy to announce the latest in our lineup of Intel-powered compute-optimized instances, the <strong>c6in</strong>, available in 9 sizes:</p>\\n<p><img src=\\"https://dev-media.amazoncloud.cn/bf5d5708dce446aeb3f459c12fec41ef_image.png\\" alt=\\"image.png\\" /></p>\n<p>The <strong>r6in</strong> and <strong>r6idn</strong> instances are available in the US East (Ohio, N. Virginia), US West (Oregon), and Europe (Ireland) regions in On-Demand and Spot form. [Savings Plans](https://aws.amazon.com/cn/savingsplans/?trk=cndc-detail) and Reserved Instances are available.</p>\\n<h3><a id=\\"Inside_the_Instances_20\\"></a><ins>Inside the Instances</ins></h3>\\n<p>As you can probably guess from these specs and from the blog post that I wrote to launch the <strong>c6in</strong> instances, all of these new instance types have a lot in common. I’ll do a rare cut-and-paste from that post in order to reiterate all of the other cool features that are available to you:</p>\\n<p><strong>Ice Lake Processors</strong> – The 3rd generation Intel Xeon Scalable processors run at 3.5 GHz, and (<a href=\\"https://edc.intel.com/content/www/us/en/products/performance/benchmarks/3rd-generation-intel-xeon-scalable-processors/\\" target=\\"_blank\\">according to Intel</a>) offer a 1.46x average performance gain over the prior generation. All-core <a href=\\"https://www.intel.com/content/www/us/en/support/articles/000007359/processors/intel-core-processors.html\\" target=\\"_blank\\">Intel Turbo Boost</a> mode is enabled on all instance sizes up to and including the <strong>12xlarg</strong>e. On the larger sizes, you can control the C-states. Intel <a href=\\"https://www.intel.com/content/www/us/en/architecture-and-technology/total-memory-encryption-security-paper.html\\" target=\\"_blank\\">Total Memory Encryption</a> (TME) is enabled, protecting instance memory with a single, transient 128-bit key generated at boot time within the processor.</p>\\n<p><strong>NUMA</strong> – Short for Non-Uniform Memory Access, this important architectural feature gives you the power to optimize for workloads where the majority of requests for a particular block of memory come from one of the processors, and that block is “closer” (architecturally speaking) to one of the processors. You can control processor affinity (and take advantage of NUMA) on the <strong>24xlarge</strong> and <strong>32xlarge</strong> instances.</p>\\n<p><strong>Networking</strong> – <a href=\\"https://aws.amazon.com/about-aws/whats-new/2016/06/introducing-elastic-network-adapter-ena-the-next-generation-network-interface-for-ec2-instances/\\" target=\\"_blank\\">Elastic Network Adapter</a> (ENA) is available on all sizes of <strong>m6in, m6idn, c6in, r6in</strong>, and <strong>r6idn</strong> instances, and <a href=\\"https://aws.amazon.com/hpc/efa/\\" target=\\"_blank\\">Elastic Fabric Adapter</a> (EFA) is available on the <strong>32xlarge</strong> instances. In order to make use of these adapters, you will need to make sure that your AMI includes the latest NVMe and ENA drivers. You can also make use of <a href=\\"https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/placement-groups.html\\" target=\\"_blank\\">Cluster Placement Groups</a>.</p>\\n<p><strong>io2 Block Express</strong> – You can use all types of EBS volumes with these instances, including the io2 Block Express volumes that we launched earlier this year. As Channy shared in his post (<a href=\\"https://aws.amazon.com/blogs/aws/amazon-ebs-io2-block-express-volumes-with-amazon-ec2-r5b-instances-are-now-generally-available/\\" target=\\"_blank\\">Amazon EBS io2 Block Express Volumes with Amazon EC2 R5b Instances Are Now Generally Available</a>), these volumes can be as large as 64 TiB, and can deliver up to 256,000 IOPS. As you can see from the tables above, you can use a <strong>24xlarge</strong> or <strong>32xlarge</strong> instance to achieve this level of performance.</p>\\n<h3><a id=\\"Choosing_the_Right_Instance_32\\"></a><ins>Choosing the Right Instance</ins></h3>\\n<p>Prior to today’s launch, you could choose a <strong>c5n, m5n</strong>, or <strong>r5n</strong> instance to get the highest network bandwidth on an EC2 instance, or an <strong>r5b</strong> instance to have access to the highest EBS IOPS performance and high EBS bandwidth. Now, customers who need high networking or EBS performance can choose from a full portfolio of instances with different memory to vCPU ratio and instance storage options available, by selecting one of <strong>c6in, m6in, m6idn, r6in</strong>, or <strong>r6idn</strong> instances.</p>\\n<p>The higher performance of the** c6in** instances will allow you to scale your network intensive workloads that need a low memory to vCPU, such as network virtual appliances, caching servers, and gaming hosts.</p>\n<p>The higher performance of <strong>m6in</strong> instances will allow you to scale your network and/or EBS intensive workloads such as data analytics, and telco applications including 5G User Plane Functions (UPF). You have the option to use the <strong>m6idn</strong> instance for workloads that benefit from low-latency local storage, such as high-performance file systems, or distributed web-scale in-memory caches.</p>\\n<p>Similarly, the higher network and EBS performance of the <strong>r6in</strong> instances will allow you to scale your network-intensive SQL, NoSQL, and in-memory database workloads, with the option to use the <strong>r6idn</strong> when you need low-latency local storage.</p>\\n<p>— <a href=\\"https://twitter.com/jeffbarr\\" target=\\"_blank\\">Jeff</a>;</p>\\n<p><img src=\\"https://dev-media.amazoncloud.cn/f3961be7c6684707af3ad9efb9a3843d_image.png\\" alt=\\"image.png\\" /></p>\n<h3><a id=\\"Jeff_Barrhttpsawsamazoncomblogsawsauthorjbarr_45\\"></a><strong><a href=\\"https://aws.amazon.com/blogs/aws/author/jbarr/\\" target=\\"_blank\\">Jeff Barr</a></strong></h3>\n<p>Jeff Barr is Chief Evangelist for Amazon Web Services. He started this blog in 2004 and has been writing posts just about non-stop ever since.</p>\n"}