This content is part of the Conference Coverage: A conference guide to AWS re:Invent 2022

Conference Coverage

Browse Sections

AWS rolls out new EC2 instances at re:Invent 2022

The wrong instance type can affect workload performance and even increase costs. This year at re:Invent, AWS released new EC2 instances from general purpose to HPC optimized.

AWS re:Invent had numerous announcements, including new additions to Amazon EC2. Cloud teams can look forward to new Amazon EC2 instances, such as general purpose, compute optimized and memory optimized options. Also, there are some Intel-powered instances in preview and soon-to-be-released to keep an eye out for.

General purpose instances

M6in and M6idn instances. These instances are for network-intensive workloads, such as application development environments, real-time big data analytics and gaming servers. They provide higher memory bandwidth per vCPU, networking speed and Amazon Elastic Block Store (EBS) bandwidth than M5n and M5dn instances. Currently both have nine sizes that have up to 200 Gbps of network bandwidth and up to 80 Gbps of EBS bandwidth. They are powered by third-generation Intel Xeon Scalable processors that run at 3.5 GHz.

C6in instances. These instances are meant for network-intensive workloads, such as distributed computing applications, data analytics and high-performance computing (HPC). They provide more network -- up to 200 Gbps -- and EBS bandwidth -- up to 80 Gbps -- than C5n instances. These nine sizes use third-generation Intel Xeon Scalable processors that run at 3.5 GHz, as well as the AWS Nitro System.

Compute-optimized instances

C7gn instances. Currently in preview, these are best suited for network-intensive workloads, such as load balancers, tightly coupled cluster computing jobs and CPU-based AI/ML inference. They are powered by Arm-based AWS Graviton3E processors and the AWS Nitro System. They provide up to 200 Gbps of networking bandwidth and up to 25 Gbps EBS bandwidth. There will be multiple sizes with up to 64 vCPUs and 128 GiB of memory.

Memory-optimized instances

R6in and R6idn instances. These are for workloads that require high networking throughput, such as processing large data sets. This is common for databases, including SAP, SQL and NoSQL, as well as SAP HANA. They provide higher memory bandwidth per vCPU, networking speed -- up to 200 Gbps -- and EBS bandwidth -- up to 80 Gbps -- than R5n and R5dn instances. They use third-generation Intel Xeon Scalable processors that run at 3.5 GHz.

R7iz instances. Currently in preview, these are powered by fourth-generation Intel Xeon Scalable processors that run at 3.9 GHz and built on the AWS Nitro System. They provide up to 50 Gbps network speed and 40 Gbps EBS bandwidth. They are a good option for simulation workloads, as well as workloads for relational databases and other commercial software licensed on a per-core basis. They will have multiple sizes with up to 128 vCPUs and 1 TiB of memory.

HPC-optimized instances

Hpc7g instances. Currently in preview and powered by AWS Graviton3E processors, these are designed for tightly coupled compute-intensive HPC and distributed computing workloads. They have up to 200 Gbps network bandwidth, which is optimized for traffic between instances in the same VPC. They will have multiple sizes with 64 vCPUs and 128 GiB of memory.

Hpc6id instances. These instances have 64 cores of Intel Xeon Scalable processors that run at 3.5 GHz with 5 GB/s per vCPU of memory bandwidth and 1024 GiB of memory. They are best suited for tightly coupled HPC workloads such as finite element analysis for various types of simulations. They have a network bandwidth of 25 and an Elastic Fabric Adapter of up to 200 Gbps.

Accelerated computing instances

Inf2 instances. Currently in preview, these are meant for deep learning (DL) inference applications, such as language translation, speech recognition, personalization and fraud detection. They are powered by up to 12 AWS Inferentia2 accelerators and offer up to 2.3 petaflops of DL performance, up to 384 GB of accelerator memory with 9.8 TB/s of memory bandwidth and up to 100 Gbps networking. There are currently four sizes.

Next Steps

Best practices for right-sizing EC2 instances

Dig Deeper on Cloud infrastructure design and management

Data Center
ITOperations
SearchAWS
SearchVMware
Close