Title here
Summary here
AWS EC2 Instance Types
Welcome back to the second article of the AWS Series! In this article, we will discuss about the different types of AWS EC2 instances that are currently offered by AWS. We will also discuss about their configurations and use cases.
We will look into the following topics in this article.
Relationship between Instance size and vCPU
It is important to understand what vCPU is, before jumping to the different categories of AWS EC2 instances. As we know, AWS EC2 instances support multithreading, which enables multiple threads to concurrently run on a single CPU core. Here, each thread is represented by a virtual CPU (vCPU) instance. In AWS terminology, “each vCPU is a thread of a CPU core, except for T2 instances”. Below table shows the common relationship between different instance EC2 instance sizes and vCPU.
Instance Size Vs vCPU
| Instance Size | vCPU |
|---|---|
| nano, micro, small, medium | 1 |
| large | 2 |
| xlarge | 4 |
| 2xlarge | 8 |
| 3xlarge | 12 |
| 4xlarge | 16 |
| 6xlarge | 24 |
| 8xlarge | 32 |
| 9xlarge | 36 |
| 10xlarge | 40 |
| 12xlarge | 48 |
| 16xlarge | 64 |
| 18xlarge | 72 |
| 24xlarge | 96 |
| 32xlarge | 128 |
Generally, every instance has a default number of CPU cores and this number varies with instance type. Also, to suit specific workloads, AWS provides an option to customize the following CPU options while launching the instance:
- Number of CPU cores for the instance.
- Disable multithreading by specifying a single thread per CPU core.
The catch is that, there is no additional charge for specifying CPU options. This means, the cost is as same as launching an EC2 instance with default CPU options.
Types of AWS EC2 Instances
AWS EC2 instances are broadly classified into 5 major categories. Each type is optimized for specific set of applications. They are as below:
- General Purpose
- Compute Optimized
- Memory Optimized
- Accelerated Computing
- Storage Optimized
Now, let’s take a look at each of these types in detail.
Click here to learn about launching an EC2 instance using AWS management console.
General Purpose Instance Type
- As the name implies, they provide a balance of compute, memory and networking resources.
- In other words, it can be used for a variety of workloads.
- It is ideal for apps that uses resources in equal proportions. For example, web server and code repos.
- EBS optimized by default, with no additional cost.
- Also, support for Enhanced Networking using Elastic Network Adapter (ENA) is available.
- For applications which requires direct access to physical processor and memory, bare metal instance sized is available.
The below table describes the types of general purpose instances and their use cases.
| Category | Processor | Features | Use cases | Storage type | Instance types |
|---|---|---|---|---|---|
| A1(a1) | custom built AWS Graviton Processor with 64-bit ARM cores | Network performance upto 10Gbps | Web servers, containerized micro services, caching fleets, distributed data stores and dev environments, Scale out and Arm-based workloads | EBS Only | medium, large, xlarge, 2xlarge, 4xlarge, metal |
| T3(t3) | 2.5GHz Intel scalable processor, burstable CPU, consistent baseline performance | Network perfromance upto 5 Gbps | Micro-services, low-latency interactive apps, small and medium databases, virtual desktops, dev environments, code repositories, and business-critical apps | EBS Only | nano, micro, small, medium, large, xlarge, 2xlarge |
| T3a(t3a) | 2.5GHz AMD EPYC 7000 series processors | Network perfromance upto 5 Gbps, 2.5GHz AMD EPYC 7000 series processors, burstable CPU, consistent baseline performance | Micro-services, low-latency interactive apps, small and medium databases, virtual desktops, dev environments, code repositories, and business-critical apps | EBS Only | nano, micro, small, medium, large, xlarge, 2xlarge |
| T2(t2) | High frequency Intel Xeon processors | Low cost general purpose instance type, free tier eligible, High frequency Intel Xeon processors, Burstable CPU and consistent baseline performance | Websites and web apps, dev, test and staging environments, build servers, code repositories, micro services and LoB apps. | EBS Only | nano, micro, small, medium, large, xlarge, 2xlarge |
| M6g(m6g) | Custom built AWS Graviton2 Processor | Network perfromance upto 25 Gbps of network bandwidth,; 40% better performance than M5 | Apps built on open-source software such as application servers, microservices, gaming servers, mid-size data stores, and caching fleets. | EBS Only | medium, large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge |
| M5(m5) | Custom built AWS Graviton Processor with 64-bit ARM cores | Network performance upto 10Gbps; provides balance of compute, memory and network resources | Small and mid-size DBs, data processing tasks that require additional memory, caching fleets, backend servers for SAP, Microsoft SharePoint, cluster computing | EBS Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge,12xlarge, 16xlarge, 24xlarge, metal |
| M5(m5d) | Custom built AWS Graviton Processor with 64-bit ARM cores | Similar to M5, except that NVMe SSD type instance storage is available | Small and mid-size DBs, data processing tasks that require additional memory, caching fleets, backend servers for SAP, Microsoft SharePoint, cluster computing | NVMe-based SSD | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge, metal |
| M5a(m5a) | 2.5 GHz AMD EPYC 7000 series processors with an all core turbo clock speed of 2.5 GHz | Upto 10% cost savings over other instances | Small and mid-size databases, data processing tasks that require additional memory, caching fleets, and for running backend servers for SAP, Microsoft SharePoint, cluster computing | EBS Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| M5a(m5ad) | 2.5 GHz AMD EPYC 7000 series processors with an all core turbo clock speed of 2.5 GHz | Similar to M5a, except that NVMe SSD type instance storage is available | Small and mid-size databases, data processing tasks that require additional memory, caching fleets, and for running backend servers for SAP, Microsoft SharePoint, cluster computing | NVMe-based SSD | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| M5n(m5n) | Custom 2nd generation Intel Xeon Scalable Processors | 25Gbps of peak BW on smaller instance sizes and 100 Gbps of newtork BW on the largest instance | Web and app servers, small and mid-sized databases, cluster computing, gaming servers, caching fleets, and other enterprise applications | EBS Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| M5n(m5nd) | Custom 2nd generation Intel Xeon Scalable Processors | Similar to M5n, except that NVMe SSD type instance storage is available | Web and app servers, small and mid-sized databases, cluster computing, gaming servers, caching fleets, and other enterprise applications | NVMe-based SSD | large, xlarge, 2xlarge,4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| M4 | 2.4 GHz Intel Xeon E5-2676 v3 Processor | balance of compute, memory and network resources | Small and mid-size databases, data processing tasks that require additional memory, caching fleets, backend servers for SAP, Microsoft SharePoint, cluster computing, and other enterprise applications. | EBS Only | large, xlarge, 2xlarge, 4xlarge, 10xlarge, 16xlarge |
Compute Optimized Instance Type
- They are designed for compute bound applications which require high performance processors.
- It can be used for compute intensive applications such as media transcoding, high perfomance web servers, gaming servers, machine learning inference.
- By default, EBS is optimized.
- Also, support for Enhanced Networking using Elastic Network Adapter (ENA) is available.
- For apps which require direct access to physical processor and memory, bare metal instance sized is available.
The below table describes the types of compute optimized instances and their use cases.
| Category | Processor | Features | Use cases | Storage type | Instance types |
|---|---|---|---|---|---|
| C5(c5) | 2nd gen Intel Xeon Scalable Processors (Cascade Lake) or 1st gen Intel Xeon Platinum 8000 series (Skylake-SP) processor | can choose processors based on the instance size, requires HVM AMIs | High Performance Computing (HPC), web servers, scientific modelling, batch processing, distributed analytics, machine/deep learning inference, ad serving, highly scalable gaming apps with multiplayers, video encoding apps. | EBS-Only | large, xlarge, 2xlarge, 4xlarge, 9xlarge, 12xlarge, 16xlarge, 18xlarge, 24xlarge, metal |
| C5(c5d) | 2nd gen Intel Xeon Scalable Processors (Cascade Lake) or 1st gen Intel Xeon Platinum 8000 series (Skylake-SP) processor | Similar to C5, except that NVMe SSD type instance storage is available | High Performance Computing (HPC), web servers, scientific modelling, batch processing, distributed analytics, machine/deep learning inference, ad serving, highly scalable gaming apps with multiplayers, video encoding apps. | NVMe SSD | large, xlarge, 2xlarge, 4xlarge, 9xlarge, 12xlarge, 16xlarge, 18xlarge, 24xlarge, metal |
| C5n(c5n) | 3.0 GHz Intel Xeon Platinum Processor | c5.18xl instance supports EFA (Elastic Fabric Adapter) used for high level of inter-node communications | High Performance Computing (HPC), web servers, scientific modelling, batch processing, distributed analytics, machine/deep learning inference, ad serving, highly scalable gaming apps with multiplayers, video encoding apps. | EBS-Only | large, xlarge, 2xlarge, 4xlarge, 9xlarge, 18xlarge, metal |
| C4(c4) | 2.9 GHz Intel Xeon E5-2666 v3 Processor | cost effecitve performance | High performance front-end fleets, web-servers, batch processing, distributed analytics, science and engineering applications where high performance is required, ad serving apps, MMO gaming, and video encoding apps. | EBS-Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge |
Memory Optimized Instance Type
- It is designed to deliver fast performance for workloads which process large sets of data in memory.
- Support for Enhanced Networking using Elastic Network Adapter (ENA) is available.
- EBS optimized by default, with no additional cost.
- For apps which require direct access to physical processor and memory, bare metal instance sized is available.
The below table describes the types of memory optimized instances and their use cases.
Memory Optimized EC2 Instance Type
| Category | Processor | Features | Use cases | Storage type | Instance types |
|---|---|---|---|---|---|
| R5(r5) | Up to 3.1 GHz Intel Xeon Platinum Processor | Up to 3.1 GHz Intel Xeon Platinum Processor, Up to 768 GiB of memory per instance, 10% price per GiB improvement and a ~20% increased CPU performance over R4 | High performance DBs, distributed web scale in-memory caches, mid-sized in-memory DB, real time big data analytics and enterprise apps | EBS-Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge, metal |
| R5(r5d) | Up to 3.1 GHz Intel Xeon Platinum Processor | similar to r5, except that NVMe SSD type instance storage is available | High performance DBs, distributed web scale in-memory caches, mid-sized in-memory DB, real time big data analytics and enterprise apps | NVMe SSD | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge, metal |
| R5a(r5a) | 2.5 GHz AMD EPYC 7000 series processors | up to 10% lower cost per GiB memory over comparable instances, upto 20Gbps network BW and Upto 768 GiB of memory per instance | High performance DBs, distributed web scale in-memory caches, mid-sized in-memory DB, real time big data analytics and enterprise apps | EBS-Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| R5a(r5ad) | 2.5 GHz AMD EPYC 7000 series processors | similar to r5a, except that NVMe SSD type instance storage is available | High performance DBs, distributed web scale in-memory caches, mid-sized in-memory DB, real time big data analytics and enterprise apps | NVMe SSD | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| R5n(r5n) | Custom 2nd generation Intel Xeon Scalable Processors (Cascade Lake) | All-core Turbo CPU frequency of 3.1 GHz and maximum single core turbo frequency of 3.5 GHz, improved network throughput and packet rate performance. | High performance DBs, distributed web scale in-memory caches, mid-sized in-memory DB, real time big data analytics and enterprise apps | EBS-Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| R5n(r5nd) | Custom 2nd generation Intel Xeon Scalable Processors (Cascade Lake) | similar to r5n, except that NVMe SSD type instance storage is available | High performance DBs, distributed web scale in-memory caches, mid-sized in-memory DB, real time big data analytics and enterprise apps | NVMe SSD | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 12xlarge, 16xlarge, 24xlarge |
| R4(r4) | 2.3 GHz Intel Xeon E5-2686 v4 Processor | Offer better price per GiB of RAM than R3, DDR4 memory | High performance DB, data mining & analysis, in-memory DB, distributed web scale in-memory caches, Real-time processing of unstructured big data, Hadoop/Spark clusters. | EBS-Only | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 16xlarge |
| X1e(x1e) | 2.3 GHz Intel Xeon E7-8880 v3 Processor | One of the lowest price per GiB of RAM, Up to 3,904 GiB of DRAM-based instance memory | High performance DBs, in-memory DBs and memory intensive apps. 32xlarge instance certified by SAP to run next-gen Business Suite S/4HANA, Business Suite on HANA (SoH), Business Warehouse on HANA (BW), and Data Mart Solutions on HANA on the AWS cloud. | EBS and SSD | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 16xlarge, 32xlarge |
| X1(x1_) | 2.3 GHz Intel Xeon E7-8880 v3 Processor | One of the lowest price per GiB of RAM, Up to 1,952 GiB of DRAM-based instance memory | In-memory DBs (e.g. SAP HANA), big data processing engines like Apache Spark, high performance computing. Certified by SAP to run Business Warehouse on HANA (BW), Data Mart Solutions on HANA, Business Suite on HANA (SoH), Business Suite S/4HANA. | EBS and SSD | 16xlarge, 32xlarge |
| High Memory (u-) | 2.1 GHz Intel Xeon Platinum 8176M processor for 6,9 and 12TB instances and 2nd Gen 2.7 GHz Intel Xeon Scalable processor for 18 and 24TB instances | 6, 9, 12, 18, and 24 TiB of instance memory, the largest of any EC2 instance, built to run large in-memory DBs, Bare metal performance | Large enterprise DBs, including prod installations of SAP HANA in-memory DB in the cloud. Certified by SAP for running Business Suite on HANA, the next-generation Business Suite S/4HANA, Data Mart Solutions on HANA, Business Warehouse on HANA, and SAP BW/4HANA in prod environments. | EBS-Only | 6tb1.metal, 9tb1.metal, 12tb1.metal, 18tb1.metal, 24tb1.metal |
| z1d(z1d) | Up to 4.0 GHz Intel® Xeon® Scalable Processors | Fastest core frequency of any cloud instance upto 4GHz,offer both high compute capacity and a high memory footprint | Ideal for electronic design automation (EDA) and certain relational DB workloads with high per-core licensing costs. | NVMe SSD | large, xlarge, 2xlarge, 3xlarge, 6xlarge, 12xlarge, metal |
Accelerated Computing Instance Type
- They are used to perform complex computing in a very short duration of time.
- In order to achieve this, they use hardware accelerators or co-processors.
- It is used for applications which has complex functions such as floating point number calculations, graphics processing or data pattern matching.
- Also, support for Enhanced Networking using Elastic Network Adapter (ENA) is available.
The below table describes the types of accelerated computing instances and their use cases.
Accelerated EC2 Instance Type
| Category | Processor | Features | Use cases | Storage type | Instance types |
|---|---|---|---|---|---|
| P3(p3) | High frequency Intel Xeon E5-2686 v4 (Broadwell) | Up to 8 NVIDIA Tesla V100 GPUs, each pairing 5,120 CUDA Cores and 640 Tensor Cores | Machine/Deep learning, high performance computing, computational fluid dynamics, computational finance, seismic analysis, speech recognition, autonomous vehicles, drug discovery. | EBS-Only | p3.2xlarge, p3.8xlarge, p3.16xlarge |
| P3(p3dn) | 2.5 GHz (base) Intel Xeon P-8175M processors | similar to p3, except that High frequency 2.5 GHz (base) Intel Xeon P-8175M processors are used | Machine/Deep learning, high performance computing, computational fluid dynamics, computational finance, seismic analysis, speech recognition, autonomous vehicles, drug discovery. | NVMe SSD | 42xlarge |
| P2(p2) | 2.3 GHz (base) and 2.7 GHz (turbo) Intel Xeon E5-2686 v4 Processor | High-performance NVIDIA K80 GPUs, each with 2,496 parallel processing cores and 12GiB of GPU memory | Machine learning, high performance DBs, computational fluid dynamics, computational finance, seismic analysis, molecular modeling, genomics, rendering, and other GPU compute workloads. | EBS-Only | xlarge, 8xlarge, 16xlarge |
| Inf1(inf1) | High frequency 2nd Gen Intel Xeon Scalable processors | AWS Neuron SDK, Up to 16 AWS Inferentia Chips, Up to 100 Gbps networking | Recommendation engines, forecasting, analysis of image and video, advanced text analytics, document analysis, voice, conversational agents, translation, transcription, and fraud detection. | EBS-Only | xlarge, 2xlarge, 6xlarge, 24xlarge |
| G4(g4) | 2.5 GHz Cascade Lake 24C processors | Up to 1.8 TB of local NVMe storage, Up to 100 Gbps of networking throughput | Machine learning inference for apps like adding metadata to an image, object detection, recommender systems, automated speech recognition, and language translation. Cost-effective platform for building and running graphics-intensive apps, such as remote graphics workstations, video transcoding, photo-realistic design, and game streaming in the cloud. | NVMe SSD | xlarge, 2xlarge, 4xlarge, 8xlarge, 16xlarge |
| G4(g4) | 2.5 GHz Cascade Lake 24C processors | Up to 1.8 TB of local NVMe storage, Up to 100 Gbps of networking throughput | Machine learning inference for apps like adding metadata to an image, object detection, recommender systems, automated speech recognition, and language translation. Cost-effective platform for building and running graphics-intensive apps, such as remote graphics workstations, video transcoding, photo-realistic design, and game streaming in the cloud. | NVMe SSD | 12xlarge, metal |
| G3(g3) | 2.3 GHz (base) and 2.7 GHz (turbo) Intel Xeon E5-2686 v4 Processor | NVIDIA Tesla M60 GPUs, each with 2048 parallel processing cores and 8 GiB of video memory | 3D visualizations and rendering, graphics-intensive remote workstation, Streaming apps, video encoding, and Graphic intensive workloads on the server side. | NVMe SSD | xlarge, 4xlarge, 8xlarge, 16xlarge |
| F1(f1) | 2.3 GHz (base) and 2.7 GHz (turbo) Intel Xeon E5-2686 v4 Processor | Approx 2.5 million logic elements and 6,800 Digital Signal Processing engines | Genomics research, financial analytics, real-time video processing, big data analysis and security. | NVMe SSD | 2xlarge, 4xlarge, 16xlarge |
Storage Optimized Instance Type
- They are mainly used for storing huge volume of date sets in a optimized manner.
- Therefore, it can be used for workloads which requires high sequential read and write access those large data.
- Futhermore, it is designed to deliver very low latency, high throughput and also high IOPS (Random I/O operations per second).
- EBS optimized by default, with no additional cost.
- Also, support for Enhanced Networking using Elastic Network Adapter (ENA) is available.
- For apps which require direct access to physical processor and memory, bare metal instance size is available.
The below table describes the types of storage optimized instances and their use cases.
| Category | Processor | Features | Use cases | Storage type | Instance types |
|---|---|---|---|---|---|
| I3(i3) | 2.3 GHz Intel Xeon E5 2686 v4 Processor | Upto 25Gbps of network BW; High Random I/O perf and High Sequential Read throughput | NoSQL DBs, in-memory DBs, scale-out transactional DBs, DW ( Data warehouse), Elasticsearch, analytics workloads. | NVMe-based SSD | large, xlarge, 2xlarge, 4xlarge, 8xlarge, 16xlarge, metal |
| I3en(i3en) | 3.1 GHz all core turbo Intel Xeon Scalable (Skylake) processors | Upto 100Gbps of network BW; High Random I/O perf and High Sequential Read throughput | NoSQL DBs, in-memory DBs, scale-out transactional DBs, DW, Elasticsearch, analytics workloads, DFS (distributed file systems) | NVMe-based SSD | large, xlarge, 2xlarge, 3xlarge, 6xlarge, 12xlarge, 24xlarge, metal |
| D2(d2) | 2.4 GHz Intel Xeon E5-2676 v3 Processor | Consistent high performance at launch time; High disk throughput | Massively Parallel Processing (MPP) data warehousing, distributed computing like MapReduce and Hadoop, DFS, NFS (Network file systems), log and data processing apps | HDD | xlarge, 2xlarge, 4xlarge, 8xlarge |
| H1(h1) | 2.4 GHz Intel Xeon E5-2676 v3 Processor | High disk throughput with Enhanced networking upto 25Gbps | MapReduce systems, DFS suchs HDFS, NFS, log and data processing apps and Big Data workloads | HDD | 2xlarge, 4xlarge, 8xlarge, 16xlar |
Conclusion
We have now reached the end of the article. We hope this article was informative and useful. You can also find more information about AWS EC2 instance types and their use cases here.
Please share your thoughts, questions, and suggestions in the comment section below. Do subscribe to our blog to stay up to date on any new articles.