Amazon EC2 Instance Types Comparison

Amazon EC2 Instance Types

Amazon EC2 delivers a wide variety of instance types that are optimized to accommodate multiple use cases. The instance types remain flexible by providing varying CPU, memory, storage, and networking capacity, allowing you to customize your resources depending on your applications.

Note: Each instance type has multiple instance sizes, which allow you to scale resources according to your workload requirements.

General Purpose Instance Types

General-purpose instances deliver computing, memory, and networking resources in a balance to facilitate diverse workload requirements. These instances are suitable for proportional resource applications such as code repositories and web servers.

Everyday use cases for general purpose instances include Virtual Desktops, low-latency interactive applications, Micro-Services small and medium databases, Development Environments, Code Repositories, and Business-Critical Apps.

Here are the general-purpose instances you must know:

Mac Instance Types

This EC2 instance gives developers access to macOS resources to develop, build, test, and sign applications that require the Xcode IDE. Mac instances are created on the AWS Nitro System and get facilitated by Apple Mac mini computers.

Amazon Elastic Block Store (EBS) facilitates storage and dedicated bare-metal instances accessible as dedicated hosts.

T4g Instance Type

EC2 T4g instances are delivered using Arm-based AWS Graviton2 processors and improve price utility by up to 40%. It provides various general-purpose burstable workloads facilitated by CPU Credits and consistent activity.

T3 Instance Types

Amazon EC2 T3 are burstable general-purpose instances providing a service line CPU performance level with the capacity to burst CPU usage anytime it is needed. Unlimited. T3 Instance Types are powered by AWS Nitro System, dedicated hardware, lightweight hypervisor, and high-frequency Intel Xeon Scalable processors, improving price-performance up to 30%.

T3a Instance Types

T3a instances are also burstable and deliver CPU performance to burst CPU usage anytime required. They supply a balance of resources designed for moderate CPU usage with temporary spikes. They can deliver up to 10% cost savings.

T2 Instance Types

Amazon T2 EC2 instances deliver burstable performance with service line CPU power that can burst above the baseline as long as needed. Although the instance needs higher CPU capacity for a prolonged period, there may be an additional fee of 5 cents per vCPU-hour.

M5 Instance Types

M5 EC2 instances are next-generation general-purpose instances facilitated by Intel Xeon Platinum 8175M processors to deliver a balance of computing, memory, and network resources suitable for various applications.

M6g Instance Types

Amazon EC2 M6g instances get supplied using Arm-based AWS Graviton2 processors. They can promise up to 40% better price/performance over M5 instances, with compute, memory, and networking resources different workloads.

M5a Instance Types

M5a instances get powered by AMD EPYC 7000 series processors, which deliver up to 10% cost savings over comparable instance types. NVMe-based SSDs are physically connected to your host server to provide block-level storage for the instance’s lifetime.

Common uses for M5a instances include data processing tasks requiring additional memory, cluster computing, Microsoft SharePoint, and running backend servers for SAP, caching fleets, and other enterprise applications.

M5n Instance Types

M5 instances optimize workloads that need a balance of computing, memory, and networking resources such as servers for web and applications, small to mid-size databases, gaming servers, and computing clusters. High bandwidth, M5n, and M5dn instance variants are ideal for applications that exploit packet rate performance and improve network throughput.

M5n Instance Types

Amazon M5zn instances deliver fast Intel Xeon Scalable processors through the cloud at an all-core turbo frequency of up to 4.5 GHz, with high throughput, and low latency networking.

M4 Instance Types

Amazon M4 instances deliver an equilibrium of computing, memory, and network resources, and it is a good choice for many applications. Especially data processing tasks that require additional memory, small and mid-size databases, running backend servers for SAP, Cluster Computing, Microsoft SharePoint, and many other enterprise applications.

A1 Instance Types

Amazon EC2 A1 instances help increase cost-saving and work best with scale-out and Arm-based workloads facilitated by the extensive Arm ecosystem. Also, they are the first EC2 instances that AWS Graviton Processors power with 64-bit Arm Neoverse cores designed by AWS.

Compute Optimized Instances

Elastic Cloud Compute Optimized instances increase the efficiency of compute-bound applications that require high-performance processing power. These instances help improve high-performance web servers, batch processing workloads, scientific modeling, and media transcoding. It also enhances High-performance computing (HPC), dedicated gaming and ad server engines, machine learning inference, and various intensive compute applications.

Amazon C6g Instances

Amazon C6g EC2 instances are delivered using Arm-based AWS Graviton2 processors, which provide up to 40% increased price performance over current-generation C5 instances for compute-intensive applications. Common uses for C6g instances include high-performance computing (HPC), CPU-based machine learning inference, video encoding, gaming, ad serving, batch processing, scientific modeling, and distributed analytics. 

Amazon C6gn Instance

Amazon EC2 C6gn instances are powered using Arm-based AWS Graviton2 processors, which provide up to 40% higher cost-efficiency. 

Amazon C5 Instance

Amazon C5 instances are optimized to compute-intensive workloads cost-effectively at high performance and a low price per compute ratio.

Amazon C5a Instance

C5a instances deliver leading x86 price-performance for a variety of intensive computing workloads. Everyday use cases for C5a instances include high CPU and memory capacity, distributed analytics, data transformations, batch processing, gaming, log analysis, and web applications.

Amazon C5n Instance

Amazon C5n instances are practical with High-Performance Computing (HPC) application workloads, data lakes, and network appliances, including firewalls and routers. It improves network throughput and packet rate performance efficiently.

Amazon C4 Instance

Amazon C4 instances are optimized to efficiently compute-intensive workloads at high performance, and cost-effectively.

Memory-Optimized Instances

Amazon’s Memory-optimized instances are structured to facilitate fast performance and processing for large data sets in workload memory. The best use cases for such instances include Memory-intensive applications like real-time big data analytics, open-source databases, and in-memory caches.

Amazon EC2 R6g Instances

Amazon EC2 R6g instances run on Arm-based AWS Graviton2 processors that can improve the price-performance over R5 instances by up to 40% for memory-intensive applications.

Amazon R5 EC2 Instances

EC2 R5 instances boost memory by an additional 5% per vCPU. The most considerable amount of memory you can get is 768 GiB. 

Amazon R5a Instances

R5a Memory Optimized instances enhance memory-bound workloads. They are powered using AMD EPYC 7000 series processors, which reduce memory costs per GiB up to 10%.

Amazon R5b Instances

Amazon R5b instances are EBS-optimized varieties of memory-optimized R5 instances that increase EBS performance by 3x in contrast to R5 instances.

Amazon R5 Instances

EC2 R5 instances optimize memory-bound workloads, including high-performance databases, real-time big data analytics, and mid-sized in-memory database. And web-scale in-memory cache applications that look to exploit improved network throughput and packet rate performance.

Amazon R4 Instances

R4 EC2 instances manage memory-intensive applications and offer a better price per GiB of RAM for high-performance databases. And in-memory databases, data mining & analysis, applications performing real-time processing of big unstructured data, distributed web scale in-memory caches, and Spark clusters.

Amazon EC2 X2gd Instances

Amazon X2gd EC2 instances are powered using Arm-based AWS Graviton2 processors. You get meager costs per GiB for memory-intensive workloads like open-source databases (MySQL, MariaDB, and PostgreSQL), electronic design automation (EDA) workloads, real-time caching servers, and in-memory caches (Redis, KeyDB, Memcached), and real-time analytics. Also, they improve price performance by up to 55%.

Amazon  X1e Instances

EC2 X1e instances are optimized to manage memory-intensive applications for enterprises, high-performance databases, and in-memory databases by delivering lower prices per GiB of RAM than other Amazon instance types.

Amazon X1e Instances

EC2 X1 instances are optimized to manage enterprise in-memory applications on a large scale with the lowest prices per GiB of RAM. Also, SAP did certify X1e instances to run Data Mart Solutions on HANA, Business Warehouse on HANA (BW), Business Suite on HANA (SoH), and Business Suite S/4HANA SAP.

Amazon High Memory Instances

EC2 High Memory instances are purpose-built to run large in-memory databases, including production deployments of SAP HANA, in the cloud.

Amazon z1d Instances

Amazon EC2 z1d instances deliver high compute capacity and memory footprint sustaining all core frequencies of up to 4.0 GHz, which is the fastest of cloud instances.

Accelerated Computing Instances

Accelerated computing instances get facilitated using hardware accelerators to perform functions such as graphics processing and data matching more efficiently than software running on CPUs.

These instances get widely used to improve machine learning. But they also work best for drug discovery, high-performance computing, seismic analysis, autonomous vehicles, computational finance, computational fluid dynamics, and speech recognition.

Amazon P4 Instances

Amazon P4 instances are GPU-based and provide top-quality performance for machine learning training and high-performance computing in the cloud.

Amazon P3 Instances

Amazon EC2 P3 instances can use up to 8 NVIDIA V100 Tensor Core GPUs to increase cloud computing performance. They can also deliver up to 100 Gbps of networking throughput to enhance machine learning and high-performance computing applications.

Amazon EC2 P2 Instances

EC2 P2 instances work with general-purpose GPU computing applications, and they support peer-to-peer GPU interactions. Also, they enhance networking using Elastic Network Adapters (ENA) with up to 25 Gbps of aggregate network capacity in Placement Groups

Amazon EC2 Inf1 Instances

Amazon EC2 Inf1 instances are foundational, and they support machine learning inference applications. They are suitable for forecasting applications, image and video analysis, recommendation engines, conversational agents, advanced text analytics, document analysis, voice, transcription and translation, and fraud detection.

Amazon G4dn Instance Types

G4dn instances help accelerate machine learning inference and graphics-intensive workloads. G4 instances provide a cost-efficient platform for building and running graphics-intensive applications like remote workstations, game cloud streaming, video transcoding, photo-realistic design, etc. 

Amazon G4ad Instances

G4ad instances provide quality price performance for graphics-intensive applications running in the cloud: game streaming, remote graphics workstations, photo-realistic design, and video transcoding. 

Amazon G3 Instances

G3 EC2 instances run graphics-intensive applications, including server-side graphics workloads, 3D visualizations, 3D rendering, remote workstation, video encoding, and application streaming. 

Amazon Ec2 F1 Instances

EC2 F1 instances offer customizable hardware acceleration instances with field-programmable gate arrays (FPGAs). They are best suited for applications that run financial analytics, genomics research, big data search and analysis, security, and real-time video processing.

Storage Optimized Instances

Storage optimized instances manage workloads that require excessive read and write access to large data sets stored locally. Optimization ensures the delivery of multiple low-latency, random I/O operations per second (IOPS) to warranted applications.

Common uses for storage optimized EC2 instances include NoSQL databases (e.g., Cassandra, MongoDB, Redis), scale-out transactional databases, Elasticsearch, and in-memory databases (e.g., Aerospike), data warehousing, and analytics workloads.

Amazon I3 Instance

EC2 I3 instances provide Non-Volatile Memory Express (NVMe) SSD-backed instance storage that delivers low latency, high sequential read throughput, and high random I/O performance at a low cost.

Amazon I3en Instances

Amazon I3en delivers Non-Volatile Memory Express (NVMe) SSD instance storage optimized to reduce latency and increase sequential disk throughput at the lowest price per GB of SSD instance storage.

I3en instances also provide bare metal instances (i3en.metal) for non-virtualized workloads that require access to physical resources for benefits. Or even workloads with possible license restrictions.

Amazon D2 Instance Type

EC2 D2 instances contain up to 48 TB of HDD-based local storage and deliver high disk throughput at the lowest price per disk throughput performance on Amazon.

Everyday use cases for D2 instances include network file systems, Massive Parallel Processing, data warehousing, and log or data-processing applications. MapReduce, Hadoop distributed computing and distributed file systems.

Amazon D3 Instances

Amazon D3 EC2 instances best manage high sequential I/O performance and disk throughput applications.

Amazon D3en Instances

Amazon D3en instances optimize large data applications that need high sequential I/O performance, disk throughput, and low-cost storage. D3en instances have the lowest cloud data storage costs

Amazon H1 Instances

EC2 Amazon H1 instances provide high disk throughput, up to 16 TB of HDD-based local storage, and equilibrium of computing and memory best suited to manage network file systems, distributed file systems such as HDFS and MapR-FS, MapReduce-based workloads, log or data processing applications such as Apache Kafka, and big data workload clusters.

Measuring Instance Performance Takeaway

Amazon EC2 provisions various instance types with different CPU, memory, disk, and networking capacity. You can choose from plenty of options across ten different instance types, each with multiple size options, organized into six distinct instance families suited for various applications.

It is beneficial to measure the performance of your applications for better identification of appropriate instance types and validation of application architecture.