Memory optimized instances - Amazon Elastic Compute Cloud
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Memory optimized instances

Note

For detailed instance type specifications, see the Amazon EC2 Instance Types Guide. For pricing information, see Amazon EC2 Instance Types.

Memory optimized instances are designed to deliver fast performance for workloads that process large data sets in memory.

R5, R5a, R5b, and R5n instances

These instances are well suited for the following:

  • High-performance, including relational MySQL and NoSQL, for example MongoDB and Cassandra databases.

  • Distributed web scale cache stores that provide in-memory caching of key-value type data, for example Memcached and Redis

  • In-memory databases using optimized data storage formats and analytics for business intelligence; for example, SAP HANA

  • Applications performing real-time processing of big unstructured data, using Hadoop and Spark clusters

  • High-performance computing (HPC) and Electronic Design Automation (EDA) applications.

Bare metal instances, such as r5.metal, provide your applications with direct access to physical resources of the host server, such as processors and memory.

R6a instances

These instances are ideal for running memory-intensive workloads, such as the following:

  • High-performance databases, both relational and NoSQL

  • Distributed web scale in-memory caches, such as Memcached and Redis

  • Real-time big data analytics, such as Hadoop and Spark clusters

R6i and R6id instances

These instances are ideal for running memory-intensive workloads, such as the following:

  • High-performance databases, relational and NoSQL

  • In-memory databases, for example SAP HANA

  • Distributed web scale in-memory caches, for example Memcached and Redis

  • Real-time big data analytics, including Hadoop and Spark clusters

R6in and R6idn instances

These instances are well suited for network-intensive workloads such as the following:

  • High-performance relational, MySQL and NoSQL. For example, MongoDB and Cassandra databases

  • Distributed web scale cache stores that provide in-memory caching of key-value type data, including Memcached and Redis

  • In-memory databases using optimized data storage formats and analytics for business intelligence, for example SAP HANA

  • Real-time big data analytics for financial services, for example Hadoop and Spark clusters

For more information, see Amazon EC2 R6i Instances.

R7a instances

These instances are powered by 4th generation AMD EPYC processors and are ideal for running memory-intensive workloads, such as the following:

  • High-performance databases, both relational and NoSQL

  • Distributed web scale in-memory caches, such as Memcached and Redis

  • Real-time big data analytics, such as Hadoop and Spark clusters

R7i instances

R7i instances are powered by 4th generation Intel Xeon processors and are ideal for running memory-intensive workloads, such as the following:

  • High-performance databases

  • Distributed web scale in-memory caches

  • In-memory databases, such as SAP HANA

  • Real-time big-data analytics, such as Hadoop and Spark clusters

R7iz instances

R7iz are high-frequency and high memory instances that are powered by 4th generation Intel Xeon processors. They provide sustained all-core turbo frequency of up to 3.9 GHz, up to 1024 GB of system memory, up to 50 Gbps network bandwidth, and up to 40 Gbps of dedicated Amazon EBS bandwidth.

R7iz instances are well suited to workloads that require a combination of high compute and high memory such as the following:

  • Electronic design automation

  • Relational databases

  • Data analytics simulations

For more information, see Amazon EC2 R7iz instances.

R7g and R7gd instances

These instances are powered by Amazon Graviton3 processors and are ideal for running memory-intensive workloads, such as the following:

  • Open-source databases; for example, MySQL, MariaDB, and PostgreSQL

  • In-memory caches; for example, Memcached, Redis, and KeyDB

For more information, see Amazon EC2 R7g instances.

High memory (u-*) instances

These instances offer 3 TiB, 6 TiB, 9 TiB, 12 TiB, 18 TiB, and 24 TiB of memory per instance. They are designed to run large in-memory databases, including production deployments of the SAP HANA in-memory database.

X1 instances

These instances are well suited for the following:

  • In-memory databases such as SAP HANA, including SAP-certified support for Business Suite S/4HANA, Business Suite on HANA (SoH), Business Warehouse on HANA (BW), and Data Mart Solutions on HANA. For more information, see SAP HANA on the Amazon Cloud.

  • Big-data processing engines such as Apache Spark or Presto.

  • High-performance computing (HPC) applications.

X1e instances

These instances are well suited for the following:

  • High-performance databases.

  • In-memory databases such as SAP HANA. For more information, see SAP HANA on the Amazon Cloud.

  • Memory-intensive enterprise applications.

X2idn, X2iedn, and X2iezn instances

These instances are well suited for the following:

  • In-memory databases, such as Redis and Memcached.

  • Relational databases, such as MySQL and PostGreSQL.

  • Electronic design automation (EDA) workloads, such as physical verification and layout tools.

  • Memory-intensive workloads, such as real-time analytics and real-time caching servers.

z1d instances

These instances deliver both high compute and high memory and are well-suited for the following:

  • Electronic Design Automation (EDA)

  • Relational database workloads

z1d.metal instances provide your applications with direct access to physical resources of the host server, such as processors and memory.

Hardware specifications

The following is a summary of the hardware specifications for memory optimized instances. A virtual central processing unit (vCPU) represents a portion of the physical CPU assigned to a virtual machine (VM). For x86 instances, there are two vCPUs per core. For Graviton instances, there is one vCPU per core.

Instance type Default vCPUs Memory (GiB)
r3.large 2 15.00
r3.xlarge 4 30.50
r3.2xlarge 8 61.00
r3.4xlarge 16 122.00
r3.8xlarge 32 244.00
r4.large 2 15.25
r4.xlarge 4 30.50
r4.2xlarge 8 61.00
r4.4xlarge 16 122.00
r4.8xlarge 32 244.00
r4.16xlarge 64 488.00
r5.large 2 16.00
r5.xlarge 4 32.00
r5.2xlarge 8 64.00
r5.4xlarge 16 128.00
r5.8xlarge 32 256.00
r5.12xlarge 48 384.00
r5.16xlarge 64 512.00
r5.24xlarge 96 768.00
r5.metal 96 768.00
r5a.large 2 16.00
r5a.xlarge 4 32.00
r5a.2xlarge 8 64.00
r5a.4xlarge 16 128.00
r5a.8xlarge 32 256.00
r5a.12xlarge 48 384.00
r5a.16xlarge 64 512.00
r5a.24xlarge 96 768.00
r5ad.large 2 16.00
r5ad.xlarge 4 32.00
r5ad.2xlarge 8 64.00
r5ad.4xlarge 16 128.00
r5ad.8xlarge 32 256.00
r5ad.12xlarge 48 384.00
r5ad.16xlarge 64 512.00
r5ad.24xlarge 96 768.00
r5b.large 2 16.00
r5b.xlarge 4 32.00
r5b.2xlarge 8 64.00
r5b.4xlarge 16 128.00
r5b.8xlarge 32 256.00
r5b.12xlarge 48 384.00
r5b.16xlarge 64 512.00
r5b.24xlarge 96 768.00
r5b.metal 96 768.00
r5d.large 2 16.00
r5d.xlarge 4 32.00
r5d.2xlarge 8 64.00
r5d.4xlarge 16 128.00
r5d.8xlarge 32 256.00
r5d.12xlarge 48 384.00
r5d.16xlarge 64 512.00
r5d.24xlarge 96 768.00
r5d.metal 96 768.00
r5dn.large 2 16.00
r5dn.xlarge 4 32.00
r5dn.2xlarge 8 64.00
r5dn.4xlarge 16 128.00
r5dn.8xlarge 32 256.00
r5dn.12xlarge 48 384.00
r5dn.16xlarge 64 512.00
r5dn.24xlarge 96 768.00
r5dn.metal 96 768.00
r5n.large 2 16.00
r5n.xlarge 4 32.00
r5n.2xlarge 8 64.00
r5n.4xlarge 16 128.00
r5n.8xlarge 32 256.00
r5n.12xlarge 48 384.00
r5n.16xlarge 64 512.00
r5n.24xlarge 96 768.00
r5n.metal 96 768.00
r6a.large 2 16.00
r6a.xlarge 4 32.00
r6a.2xlarge 8 64.00
r6a.4xlarge 16 128.00
r6a.8xlarge 32 256.00
r6a.12xlarge 48 384.00
r6a.16xlarge 64 512.00
r6a.24xlarge 96 768.00
r6a.32xlarge 128 1024.00
r6a.48xlarge 192 1536.00
r6a.metal 192 1536.00
r6i.large 2 16.00
r6i.xlarge 4 32.00
r6i.2xlarge 8 64.00
r6i.4xlarge 16 128.00
r6i.8xlarge 32 256.00
r6i.12xlarge 48 384.00
r6i.16xlarge 64 512.00
r6i.24xlarge 96 768.00
r6i.32xlarge 128 1024.00
r6i.metal 128 1024.00
r6idn.large 2 16.00
r6idn.xlarge 4 32.00
r6idn.2xlarge 8 64.00
r6idn.4xlarge 16 128.00
r6idn.8xlarge 32 256.00
r6idn.12xlarge 48 384.00
r6idn.16xlarge 64 512.00
r6idn.24xlarge 96 768.00
r6idn.32xlarge 128 1024.00
r6idn.metal 128 1024.00
r6in.large 2 16.00
r6in.xlarge 4 32.00
r6in.2xlarge 8 64.00
r6in.4xlarge 16 128.00
r6in.8xlarge 32 256.00
r6in.12xlarge 48 384.00
r6in.16xlarge 64 512.00
r6in.24xlarge 96 768.00
r6in.32xlarge 128 1024.00
r6in.metal 128 1024.00
r6id.large 2 16.00
r6id.xlarge 4 32.00
r6id.2xlarge 8 64.00
r6id.4xlarge 16 128.00
r6id.8xlarge 32 256.00
r6id.12xlarge 48 384.00
r6id.16xlarge 64 512.00
r6id.24xlarge 96 768.00
r6id.32xlarge 128 1024.00
r6id.metal 128 1024.00
r7a.medium 1 8.00
r7a.large 2 16.00
r7a.xlarge 4 32.00
r7a.2xlarge 8 64.00
r7a.4xlarge 16 128.00
r7a.8xlarge 32 256.00
r7a.12xlarge 48 384.00
r7a.16xlarge 64 512.00
r7a.24xlarge 96 768.00
r7a.32xlarge 128 1024.00
r7a.48xlarge 192 1536.00
r7a.metal-48xl 192 1536.00
r7i.large 2 16.00
r7i.xlarge 4 32.00
r7i.2xlarge 8 64.00
r7i.4xlarge 16 128.00
r7i.8xlarge 32 256.00
r7i.12xlarge 48 384.00
r7i.16xlarge 64 512.00
r7i.24xlarge 96 768.00
r7i.48xlarge 192 1536.00
r7i.metal-24xl 96 768.00
r7i.metal-48xl 192 1536.00
r7iz.large 2 16.00
r7iz.xlarge 4 32.00
r7iz.2xlarge 8 64.00
r7iz.4xlarge 16 128.00
r7iz.8xlarge 32 256.00
r7iz.12xlarge 48 384.00
r7iz.16xlarge 64 512.00
r7iz.32xlarge 128 1024.00
r7iz.metal-16xl 64 512.00
r7iz.metal-32xl 128 1024.00
u-3tb1.56xlarge 224 3072.00
u-6tb1.56xlarge 224 6144.00
u-6tb1.112xlarge 448 6144.00
u-6tb1.metal 448 6144.00
u-9tb1.112xlarge 448 9216.00
u-9tb1.metal 448 9216.00
u-12tb1.112xlarge 448 12288.00
u-12tb1.metal 448 12288.00
u-18tb1.112xlarge 448 18432.00
u-18tb1.metal 448 18432.00
u-24tb1.112xlarge 448 24576.00
u-24tb1.metal 448 24576.00
x1.16xlarge 64 976.00
x1.32xlarge 128 1952.00
x2idn.16xlarge 64 1024.00
x2idn.24xlarge 96 1536.00
x2idn.32xlarge 128 2048.00
x2idn.metal 128 2048.00
x2iedn.xlarge 4 128.00
x2iedn.2xlarge 8 256.00
x2iedn.4xlarge 16 512.00
x2iedn.8xlarge 32 1024.00
x2iedn.16xlarge 64 2048.00
x2iedn.24xlarge 96 3072.00
x2iedn.32xlarge 128 4096.00
x2iedn.metal 128 4096.00
x2iezn.2xlarge 8 256.00
x2iezn.4xlarge 16 512.00
x2iezn.6xlarge 24 768.00
x2iezn.8xlarge 32 1024.00
x2iezn.12xlarge 48 1536.00
x2iezn.metal 48 1536.00
x1e.xlarge 4 122.00
x1e.2xlarge 8 244.00
x1e.4xlarge 16 488.00
x1e.8xlarge 32 976.00
x1e.16xlarge 64 1952.00
x1e.32xlarge 128 3904.00
z1d.large 2 16.00
z1d.xlarge 4 32.00
z1d.2xlarge 8 64.00
z1d.3xlarge 12 96.00
z1d.6xlarge 24 192.00
z1d.12xlarge 48 384.00
z1d.metal 48 384.00

* Each logical processor is a hyperthread on 224 cores.

The memory optimized instances use the following processors.

AMD processors
  • AMD EPYC 7000 series processors (AMD EPYC 7571): R5a, R5ad

  • 3rd generation AMD EPYC processors (AMD EPYC 7R13): R6a

  • 4th generation AMD EPYC processors (AMD EPYC 9R14): R7a

Intel processors
  • Intel Xeon Scalable processors (Haswell E7-8880 v3): X1, X1e

  • Intel Xeon Scalable processors (Broadwell E5-2686 v4): R4

  • Intel Xeon Scalable processors (Skylake 8151): z1d

  • Intel Xeon Scalable processors (Skylake 8175M or Cascade Lake 8259CL): R5, R5d

  • 2nd generation Intel Xeon Scalable processors (Cascade Lake 8259CL): R5b, R5n

  • 2nd generation Intel Xeon Scalable processors (Cascade Lake 8252C): X2iezn

  • 3rd generation Intel Xeon Scalable processors (Ice Lake 8375C): R6i, R6id, X2idn, X2iedn

  • 4th generation Intel Xeon Scalable processors (Sapphire Rapids 8488C): R7i

  • 4th generation Intel Xeon Scalable processors (Sapphire Rapids 6455B): R7iz

For detailed instance type specifications, see the Amazon EC2 Instance Types Guide. For pricing information, see Amazon EC2 Instance Types.

Memory performance

X1 instances include Intel Scalable Memory Buffers, providing 300 GiB/s of sustainable memory-read bandwidth and 140 GiB/s of sustainable memory-write bandwidth.

For more information about how much RAM can be enabled for memory optimized instances, see Hardware specifications.

Memory optimized instances have high memory and require 64-bit HVM AMIs to take advantage of that capacity. HVM AMIs provide superior performance in comparison to paravirtual (PV) AMIs on memory optimized instances. .

Instance performance

Memory optimized instances enable increased cryptographic performance through the latest Intel AES-NI feature and support Advanced Vector Extensions 2 (Intel AVX2) processor instructions to expand most integer commands to 256 bits.

Network performance

You can enable enhanced networking on supported instance types to provide lower latencies, lower network jitter, and higher packet-per-second (PPS) performance. Most applications do not consistently need a high level of network performance, but can benefit from access to increased bandwidth when they send or receive data. For more information, see Enhanced networking on Windows.

The following is a summary of network performance for memory optimized instances that support enhanced networking.

Note

Instance types indicated with a have a baseline bandwidth and can use a network I/O credit mechanism to burst beyond their baseline bandwidth on a best effort basis. For more information, see instance network bandwidth.

Instance type Network performance Enhanced networking features
r3.large Moderate Intel 82599 VF
r3.xlarge Moderate Intel 82599 VF
r3.2xlarge High Intel 82599 VF
r3.4xlarge High Intel 82599 VF
r3.8xlarge 10 Gigabit Intel 82599 VF
r4.large Up to 10 Gigabit ENA
r4.xlarge Up to 10 Gigabit ENA
r4.2xlarge Up to 10 Gigabit ENA
r4.4xlarge Up to 10 Gigabit ENA
r4.8xlarge 10 Gigabit ENA
r4.16xlarge 25 Gigabit ENA
r5.large Up to 10 Gigabit ENA
r5.xlarge Up to 10 Gigabit ENA
r5.2xlarge Up to 10 Gigabit ENA
r5.4xlarge Up to 10 Gigabit ENA
r5.8xlarge 10 Gigabit ENA
r5.12xlarge 12 Gigabit ENA
r5.16xlarge 20 Gigabit ENA
r5.24xlarge 25 Gigabit ENA
r5.metal 25 Gigabit ENA
r5a.large Up to 10 Gigabit ENA
r5a.xlarge Up to 10 Gigabit ENA
r5a.2xlarge Up to 10 Gigabit ENA
r5a.4xlarge Up to 10 Gigabit ENA
r5a.8xlarge Up to 10 Gigabit ENA
r5a.12xlarge 10 Gigabit ENA
r5a.16xlarge 12 Gigabit ENA
r5a.24xlarge 20 Gigabit ENA
r5ad.large Up to 10 Gigabit ENA
r5ad.xlarge Up to 10 Gigabit ENA
r5ad.2xlarge Up to 10 Gigabit ENA
r5ad.4xlarge Up to 10 Gigabit ENA
r5ad.8xlarge Up to 10 Gigabit ENA
r5ad.12xlarge 10 Gigabit ENA
r5ad.16xlarge 12 Gigabit ENA
r5ad.24xlarge 20 Gigabit ENA
r5b.large Up to 10 Gigabit ENA
r5b.xlarge Up to 10 Gigabit ENA
r5b.2xlarge Up to 10 Gigabit ENA
r5b.4xlarge Up to 10 Gigabit ENA
r5b.8xlarge 10 Gigabit ENA
r5b.12xlarge 12 Gigabit ENA
r5b.16xlarge 20 Gigabit ENA
r5b.24xlarge 25 Gigabit ENA
r5b.metal 25 Gigabit ENA
r5d.large Up to 10 Gigabit ENA
r5d.xlarge Up to 10 Gigabit ENA
r5d.2xlarge Up to 10 Gigabit ENA
r5d.4xlarge Up to 10 Gigabit ENA
r5d.8xlarge 10 Gigabit ENA
r5d.12xlarge 12 Gigabit ENA
r5d.16xlarge 20 Gigabit ENA
r5d.24xlarge 25 Gigabit ENA
r5d.metal 25 Gigabit ENA
r5dn.large Up to 25 Gigabit ENA
r5dn.xlarge Up to 25 Gigabit ENA
r5dn.2xlarge Up to 25 Gigabit ENA
r5dn.4xlarge Up to 25 Gigabit ENA
r5dn.8xlarge 25 Gigabit ENA
r5dn.12xlarge 50 Gigabit ENA
r5dn.16xlarge 75 Gigabit ENA
r5dn.24xlarge 100 Gigabit ENA | EFA
r5dn.metal 100 Gigabit ENA | EFA
r5n.large Up to 25 Gigabit ENA
r5n.xlarge Up to 25 Gigabit ENA
r5n.2xlarge Up to 25 Gigabit ENA
r5n.4xlarge Up to 25 Gigabit ENA
r5n.8xlarge 25 Gigabit ENA
r5n.12xlarge 50 Gigabit ENA
r5n.16xlarge 75 Gigabit ENA
r5n.24xlarge 100 Gigabit ENA | EFA
r5n.metal 100 Gigabit ENA | EFA
r6a.large Up to 12.5 Gigabit ENA
r6a.xlarge Up to 12.5 Gigabit ENA
r6a.2xlarge Up to 12.5 Gigabit ENA
r6a.4xlarge Up to 12.5 Gigabit ENA
r6a.8xlarge 12.5 Gigabit ENA
r6a.12xlarge 18.75 Gigabit ENA
r6a.16xlarge 25 Gigabit ENA
r6a.24xlarge 37.5 Gigabit ENA
r6a.32xlarge 50 Gigabit ENA
r6a.48xlarge 50 Gigabit ENA | EFA
r6a.metal 50 Gigabit ENA | EFA
r6i.large Up to 12.5 Gigabit ENA
r6i.xlarge Up to 12.5 Gigabit ENA
r6i.2xlarge Up to 12.5 Gigabit ENA
r6i.4xlarge Up to 12.5 Gigabit ENA
r6i.8xlarge 12.5 Gigabit ENA
r6i.12xlarge 18.75 Gigabit ENA
r6i.16xlarge 25 Gigabit ENA
r6i.24xlarge 37.5 Gigabit ENA
r6i.32xlarge 50 Gigabit ENA | EFA
r6i.metal 50 Gigabit ENA | EFA
r6idn.large Up to 25 Gigabit ENA
r6idn.xlarge Up to 30 Gigabit ENA
r6idn.2xlarge Up to 40 Gigabit ENA
r6idn.4xlarge Up to 50 Gigabit ENA
r6idn.8xlarge 50 Gigabit ENA
r6idn.12xlarge 75 Gigabit ENA
r6idn.16xlarge 100 Gigabit ENA
r6idn.24xlarge 150 Gigabit ENA
r6idn.32xlarge 200 Gigabit ENA | EFA
r6idn.metal 200 Gigabit ENA | EFA
r6in.large Up to 25 Gigabit ENA
r6in.xlarge Up to 30 Gigabit ENA
r6in.2xlarge Up to 40 Gigabit ENA
r6in.4xlarge Up to 50 Gigabit ENA
r6in.8xlarge 50 Gigabit ENA
r6in.12xlarge 75 Gigabit ENA
r6in.16xlarge 100 Gigabit ENA
r6in.24xlarge 150 Gigabit ENA
r6in.32xlarge 200 Gigabit ENA | EFA
r6in.metal 200 Gigabit ENA | EFA
r6id.large Up to 12.5 Gigabit ENA
r6id.xlarge Up to 12.5 Gigabit ENA
r6id.2xlarge Up to 12.5 Gigabit ENA
r6id.4xlarge Up to 12.5 Gigabit ENA
r6id.8xlarge 12.5 Gigabit ENA
r6id.12xlarge 18.75 Gigabit ENA
r6id.16xlarge 25 Gigabit ENA
r6id.24xlarge 37.5 Gigabit ENA
r6id.32xlarge 50 Gigabit ENA | EFA
r6id.metal 50 Gigabit ENA | EFA
r7a.medium Up to 12.5 Gigabit ENA
r7a.large Up to 12.5 Gigabit ENA
r7a.xlarge Up to 12.5 Gigabit ENA
r7a.2xlarge Up to 12.5 Gigabit ENA
r7a.4xlarge Up to 12.5 Gigabit ENA
r7a.8xlarge 12.5 Gigabit ENA
r7a.12xlarge 18.75 Gigabit ENA
r7a.16xlarge 25 Gigabit ENA
r7a.24xlarge 37.5 Gigabit ENA
r7a.32xlarge 50 Gigabit ENA
r7a.48xlarge 50 Gigabit ENA | EFA
r7a.metal-48xl 50 Gigabit ENA | EFA
r7i.large Up to 12.5 Gigabit ENA
r7i.xlarge Up to 12.5 Gigabit ENA
r7i.2xlarge Up to 12.5 Gigabit ENA
r7i.4xlarge Up to 12.5 Gigabit ENA
r7i.8xlarge 12.5 Gigabit ENA
r7i.12xlarge 18.75 Gigabit ENA
r7i.16xlarge 25 Gigabit ENA
r7i.24xlarge 37.5 Gigabit ENA
r7i.48xlarge 50 Gigabit ENA | EFA
r7i.metal-24xl 37.5 Gigabit ENA
r7i.metal-48xl 50 Gigabit ENA | EFA
r7iz.large Up to 12.5 Gigabit ENA
r7iz.xlarge Up to 12.5 Gigabit ENA
r7iz.2xlarge Up to 12.5 Gigabit ENA
r7iz.4xlarge Up to 12.5 Gigabit ENA
r7iz.8xlarge 12.5 Gigabit ENA
r7iz.12xlarge 25 Gigabit ENA
r7iz.16xlarge 25 Gigabit ENA
r7iz.32xlarge 50 Gigabit ENA | EFA
r7iz.metal-16xl 25 Gigabit ENA
r7iz.metal-32xl 50 Gigabit ENA | EFA
u-3tb1.56xlarge 50 Gigabit ENA
u-6tb1.56xlarge 100 Gigabit ENA
u-6tb1.112xlarge 100 Gigabit ENA
u-6tb1.metal 100 ENA
u-9tb1.112xlarge 100 Gigabit ENA
u-9tb1.metal 100 ENA
u-12tb1.112xlarge 100 Gigabit ENA
u-12tb1.metal 100 ENA
u-18tb1.112xlarge 100 Gigabit ENA
u-18tb1.metal 100 Gigabit ENA
u-24tb1.112xlarge 100 Gigabit ENA
u-24tb1.metal 100 Gigabit ENA
x1.16xlarge 10 Gigabit ENA
x1.32xlarge 25 Gigabit ENA
x2idn.16xlarge 50 Gigabit ENA
x2idn.24xlarge 75 Gigabit ENA
x2idn.32xlarge 100 Gigabit ENA | EFA
x2idn.metal 100 Gigabit ENA | EFA
x2iedn.xlarge Up to 25 Gigabit ENA
x2iedn.2xlarge Up to 25 Gigabit ENA
x2iedn.4xlarge Up to 25 Gigabit ENA
x2iedn.8xlarge 25 Gigabit ENA
x2iedn.16xlarge 50 Gigabit ENA
x2iedn.24xlarge 75 Gigabit ENA
x2iedn.32xlarge 100 Gigabit ENA | EFA
x2iedn.metal 100 Gigabit ENA | EFA
x2iezn.2xlarge Up to 25 Gigabit ENA
x2iezn.4xlarge Up to 25 Gigabit ENA
x2iezn.6xlarge 50 Gigabit ENA
x2iezn.8xlarge 75 Gigabit ENA
x2iezn.12xlarge 100 Gigabit ENA | EFA
x2iezn.metal 100 Gigabit ENA | EFA
x1e.xlarge Up to 10 Gigabit ENA
x1e.2xlarge Up to 10 Gigabit ENA
x1e.4xlarge Up to 10 Gigabit ENA
x1e.8xlarge Up to 10 Gigabit ENA
x1e.16xlarge 10 Gigabit ENA
x1e.32xlarge 25 Gigabit ENA
z1d.large Up to 10 Gigabit ENA
z1d.xlarge Up to 10 Gigabit ENA
z1d.2xlarge Up to 10 Gigabit ENA
z1d.3xlarge Up to 10 Gigabit ENA
z1d.6xlarge 12 Gigabit ENA
z1d.12xlarge 25 Gigabit ENA
z1d.metal 25 Gigabit ENA

For 32xlarge and metal instance types that support 200 Gbps, at least 2 ENIs, each attached to a different network card, are required on the instance to achieve 200 Gbps throughput. Each ENI attached to a network card can achieve a max of 170 Gbps.

u-6tb1.metal, u-9tb1.metal, and u-12tb1.metal instances launched after March 12, 2020 provide network performance of 100 Gbps. u-6tb1.metal, u-9tb1.metal, and u-12tb1.metal instances launched before March 12, 2020 mightonly provide network performance of 25 Gbps. To ensure that instances launched before March 12, 2020 have a network performance of 100 Gbps, contact your account team to upgrade your instance at no additional cost.

The following table shows the baseline and burst bandwidth for instance types that use the network I/O credit mechanism to burst beyond their baseline bandwidth.

Instance type Baseline bandwidth (Gbps) Burst bandwidth (Gbps)
r4.large 0.75 10.0
r4.xlarge 1.25 10.0
r4.2xlarge 2.5 10.0
r4.4xlarge 5.0 10.0
r5.large 0.75 10.0
r5.xlarge 1.25 10.0
r5.2xlarge 2.5 10.0
r5.4xlarge 5.0 10.0
r5a.large 0.75 10.0
r5a.xlarge 1.25 10.0
r5a.2xlarge 2.5 10.0
r5a.4xlarge 5.0 10.0
r5a.8xlarge 7.5 10.0
r5ad.large 0.75 10.0
r5ad.xlarge 1.25 10.0
r5ad.2xlarge 2.5 10.0
r5ad.4xlarge 5.0 10.0
r5ad.8xlarge 7.5 10.0
r5b.large 0.75 10.0
r5b.xlarge 1.25 10.0
r5b.2xlarge 2.5 10.0
r5b.4xlarge 5.0 10.0
r5d.large 0.75 10.0
r5d.xlarge 1.25 10.0
r5d.2xlarge 2.5 10.0
r5d.4xlarge 5.0 10.0
r5dn.large 2.1 25.0
r5dn.xlarge 4.1 25.0
r5dn.2xlarge 8.125 25.0
r5dn.4xlarge 16.25 25.0
r5n.large 2.1 25.0
r5n.xlarge 4.1 25.0
r5n.2xlarge 8.125 25.0
r5n.4xlarge 16.25 25.0
r6a.large 0.781 12.5
r6a.xlarge 1.562 12.5
r6a.2xlarge 3.125 12.5
r6a.4xlarge 6.25 12.5
r6i.large 0.781 12.5
r6i.xlarge 1.562 12.5
r6i.2xlarge 3.125 12.5
r6i.4xlarge 6.25 12.5
r6idn.large 3.125 25.0
r6idn.xlarge 6.25 30.0
r6idn.2xlarge 12.5 40.0
r6idn.4xlarge 25.0 50.0
r6in.large 3.125 25.0
r6in.xlarge 6.25 30.0
r6in.2xlarge 12.5 40.0
r6in.4xlarge 25.0 50.0
r6id.large 0.781 12.5
r6id.xlarge 1.562 12.5
r6id.2xlarge 3.125 12.5
r6id.4xlarge 6.25 12.5
r7a.medium 0.39 12.5
r7a.large 0.781 12.5
r7a.xlarge 1.562 12.5
r7a.2xlarge 3.125 12.5
r7a.4xlarge 6.25 12.5
r7i.large 0.781 12.5
r7i.xlarge 1.562 12.5
r7i.2xlarge 3.125 12.5
r7i.4xlarge 6.25 12.5
r7iz.large 0.781 12.5
r7iz.xlarge 1.562 12.5
r7iz.2xlarge 3.125 12.5
r7iz.4xlarge 6.25 12.5
x2iedn.xlarge 1.875 25.0
x2iedn.2xlarge 5.0 25.0
x2iedn.4xlarge 12.5 25.0
x2iezn.2xlarge 12.5 25.0
x2iezn.4xlarge 15.0 25.0
x1e.xlarge 0.625 10.0
x1e.2xlarge 1.25 10.0
x1e.4xlarge 2.5 10.0
x1e.8xlarge 5.0 10.0
z1d.large 0.75 10.0
z1d.xlarge 1.25 10.0
z1d.2xlarge 2.5 10.0
z1d.3xlarge 5.0 10.0

Amazon EBS I/O performance

Amazon EBS optimized instances use an optimized configuration stack and provide additional, dedicated capacity for Amazon EBS I/O. This optimization provides the best performance for your Amazon EBS volumes by minimizing contention between Amazon EBS I/O and other traffic from your instance.

For more information, see Amazon EBS–optimized instances.

SSD-based instance store volume I/O performance

If you use all the SSD-based instance store volumes available to your instance, you can get up to the IOPS (4,096 byte block size) performance listed in the following table (at queue depth saturation). Otherwise, you get lower IOPS performance.

Instance Size 100% Random Read IOPS Write IOPS
r5ad.large 30000 15000
r5ad.xlarge 59000 29000
r5ad.2xlarge 117000 57000
r5ad.4xlarge 234000 114000
r5ad.8xlarge 466666 233334
r5ad.12xlarge 700000 340000
r5ad.16xlarge 933332 466668
r5ad.24xlarge 1400000 680000
r5d.large 30000 15000
r5d.xlarge 59000 29000
r5d.2xlarge 117000 57000
r5d.4xlarge 234000 114000
r5d.8xlarge 466666 233334
r5d.12xlarge 700000 340000
r5d.16xlarge 933332 466668
r5d.24xlarge 1400000 680000
r5d.metal 1400000 680000
r5dn.large 29000 14500
r5dn.xlarge 58000 29000
r5dn.2xlarge 116000 58000
r5dn.4xlarge 232000 116000
r5dn.8xlarge 464000 232000
r5dn.12xlarge 700000 350000
r5dn.16xlarge 930000 465000
r5dn.24xlarge 1400000 700000
r5dn.metal 1400000 700000
r6idn.large 33542 16771
r6idn.xlarge 67083 33542
r6idn.2xlarge 134167 67084
r6idn.4xlarge 268333 134167
r6idn.8xlarge 536666 268334
r6idn.12xlarge 804998 402500
r6idn.16xlarge 1073332 536668
r6idn.24xlarge 1609996 805000
r6idn.32xlarge 2146664 1073336
r6idn.metal 2146664 1073336
r6id.large 33542 16771
r6id.xlarge 67083 33542
r6id.2xlarge 134167 67084
r6id.4xlarge 268333 134167
r6id.8xlarge 536666 268334
r6id.12xlarge 804998 402500
r6id.16xlarge 1073332 536668
r6id.24xlarge 1609996 805000
r6id.32xlarge 2146664 1073336
r6id.metal 2146664 1073336
x2idn.16xlarge 430000 180000
x2idn.24xlarge 645000 270000
x2idn.32xlarge 860000 360000
x2idn.metal 860000 360000
x2iedn.xlarge 26875 11250
x2iedn.2xlarge 53750 22500
x2iedn.4xlarge 107500 45000
x2iedn.8xlarge 215000 90000
x2iedn.16xlarge 430000 180000
x2iedn.24xlarge 645000 270000
x2iedn.32xlarge 860000 360000
x2iedn.metal 860000 360000
z1d.large 30000 15000
z1d.xlarge 59000 29000
z1d.2xlarge 117000 57000
z1d.3xlarge 175000 75000
z1d.6xlarge 350000 170000
z1d.12xlarge 700000 340000
z1d.metal 700000 340000

As you fill the SSD-based instance store volumes for your instance, the number of write IOPS that you can achieve decreases. This is due to the extra work the SSD controller must do to find available space, rewrite existing data, and erase unused space so that it can be rewritten. This process of garbage collection results in internal write amplification to the SSD, expressed as the ratio of SSD write operations to user write operations. This decrease in performance is even larger if the write operations are not in multiples of 4,096 bytes or not aligned to a 4,096-byte boundary. If you write a smaller amount of bytes or bytes that are not aligned, the SSD controller must read the surrounding data and store the result in a new location. This pattern results in significantly increased write amplification, increased latency, and dramatically reduced I/O performance.

SSD controllers can use several strategies to reduce the impact of write amplification. One such strategy is to reserve space in the SSD instance storage so that the controller can more efficiently manage the space available for write operations. This is called over-provisioning. The SSD-based instance store volumes provided to an instance don't have any space reserved for over-provisioning. To reduce write amplification, we recommend that you leave 10% of the volume unpartitioned so that the SSD controller can use it for over-provisioning. This decreases the storage that you can use, but increases performance even if the disk is close to full capacity.

For instance store volumes that support TRIM, you can use the TRIM command to notify the SSD controller whenever you no longer need data that you've written. This provides the controller with more free space, which can reduce write amplification and increase performance. For more information, see Instance store volume TRIM support.

High availability and reliability (X1)

X1 instances support Single Device Data Correction (SDDC +1), which detects and corrects multi-bit errors. SDDC +1 uses error checking and correction code to identify and disable a failed single DRAM device.

In addition, you can implement high availability (HA) and disaster recovery (DR) solutions to meet recovery point objective (RPO), recovery time objective (RTO), and cost requirements by leveraging Amazon CloudFormation and Recover your instance.

If you run an SAP HANA production environment, you also have the option of using HANA System Replication (HSR) on X1 instances. For more information about architecting HA and DR solutions on X1 instances, see SAP HANA on the Amazon Web Services Cloud: Quick Start Reference Deployment.

Support for vCPUs

Memory optimized instances provide a high number of vCPUs, which can cause launch issues with operating systems that have a lower vCPU limit. We strongly recommend that you use the latest AMIs when you launch memory optimized instances.

The following AMIs support launching memory optimized instances:

  • Amazon Linux 2 (HVM)

  • Amazon Linux AMI 2016.03 (HVM) or later

  • Ubuntu Server 14.04 LTS (HVM)

  • Red Hat Enterprise Linux 7.1 (HVM)

  • SUSE Linux Enterprise Server 12 SP1 (HVM)

  • Windows Server 2019

  • Windows Server 2016

  • Windows Server 2012 R2

  • Windows Server 2012

  • Windows Server 2008 R2 64-bit

  • Windows Server 2008 SP2 64-bit

Release notes

  • r6a.metal instances do not support Windows Server 2016 and earlier.

  • r7a.metal-48xl instances support only Windows Server 2019 and Windows Server 2022 AMIs. For Windows Server 2019, you must use an AMI with a release date of July 2023 or later.

  • Instances built on the Nitro System have the following requirements:

    The current Amazon Windows AMIs meet these requirements.

  • To get the best performance from your R6i instances, ensure that they have ENA driver version 2.2.3 or later. Using an ENA driver earlier than version 2.0.0 with these instances causes network interface attachment failures. The following AMIs have a compatible ENA driver.

    • Amazon Windows AMI from May 2021 or later

  • The maximum number of Amazon EBS volumes that you can attach to an instance depends on the instance type and instance size. For more information, see Instance volume limits.

  • Launching a bare metal instance boots the underlying server, which includes verifying all hardware and firmware components. This means that it can take 20 minutes from the time the instance enters the running state until it becomes available over the network.

  • To attach or detach EBS volumes or secondary network interfaces from a bare metal instance requires PCIe native hotplug support.

  • Bare metal instances use a PCI-based serial device rather than an I/O port-based serial device. The upstream Linux kernel and the latest Amazon Linux AMIs support this device. Bare metal instances also provide an ACPI SPCR table to enable the system to automatically use the PCI-based serial device. The latest Windows AMIs automatically use the PCI-based serial device.

  • You can't launch X1 instances using a Windows Server 2008 SP2 64-bit AMI, except for x1.16xlarge instances.

  • You can't launch X1e instances using a Windows Server 2008 SP2 64-bit AMI.

  • With earlier versions of the Windows Server 2008 R2 64-bit AMI, you can't launch r4.large and r4.4xlarge instances. If you experience this issue, update to the latest version of this AMI.

  • There is a limit on the total number of instances that you can launch in a Region, and there are additional limits on some instance types. For more information, see How many instances can I run in Amazon EC2? in the Amazon EC2 FAQ.