Optimizing Cost with AWS Graviton Instances

Introduction

Amazon Web Services (AWS) offers various instance types to cater to various workloads. AWS Graviton-based instances have gained attention due to their cost-effectiveness and competitive performance. In this blog, we’ll delve into the implementation of Graviton instances, compare their pricing with x86 architecture instances, and explore their computing and memory capabilities.

AWS Graviton instances

AWS Graviton instances are a family of Amazon EC2 instances powered by ARM-based processors. These instances offer competitive performance while being cost-effective. Graviton instances are particularly well-suited for disk-bound and CPU-bound workloads. Organizations looking to optimize costs without compromising performance should consider integrating Graviton instances into their infrastructure.

Features of AWS Graviton Instances

1. Cost-Effective: Graviton instances are designed to provide excellent price-performance ratios. They cost up to 20% less than comparable x86-based Amazon EC2 instances¹.
2. ARM-Based Architecture: Graviton instances utilize ARM processors, which offer a different architecture than traditional x86 CPUs.
3. Performance: While ARM processors were once associated with mobile devices, AWS Graviton instances deliver competitive performance. They are particularly well-suited for disk-bound and CPU-bound workloads².
4. Memory Optimized: Graviton instances come in various flavors, including memory-optimized options. These are ideal for applications that require substantial memory resources.

Classes of AWS Graviton Instances

1. General Purpose Instances (M7g, M7gd):

• These instances strike a balance between compute, memory, and networking.
• Ideal for general-purpose workloads such as application servers, midsize data stores, microservices, and cluster computing.
• Powered by AWS Graviton3 processors.

2. Compute-Optimized Instances (T4g):

• Best suited for burstable general-purpose workloads.
• Great for large-scale microservices, small and medium databases, virtual desktops, and business-critical applications.
• Powered by AWS Graviton2

3. Memory-Optimized Instances (M6gd, C6gd, R6gd):

• These instances cater to memory-intensive applications.
• Offer local NVMe-based SSD storage options.
• C6gnand C7gn instances, with enhanced networking and support for Elastic Fabric Adapter (EFA), are also available.

4. Accelerated Computing Instances (X2gd, G5g, Im4gn, Is4gen, I4g):

• Designed for specialized workloads that require acceleration.
• Examples include machine learning, high-performance computing (HPC), and data analytics.

Comparison with x86 Instances

Cost Comparison

• Graviton vs. x86: Graviton instances are generally more cost-effective than traditional x86 processors. AWS provides better prices for Graviton instances, making them attractive for price-conscious customers. Graviton processors offer 15-25% better price-performance for disk-bound and CPU-bound workloads².
• Spot Prices: Sometimes, ARM instances can even exceed x86 spot prices. However, in many cases, ARM instances cost significantly less than their x86 equivalents³.

Performance Comparison

• Computational Performance: AWS Graviton instances perform similarly to x86_64 instances. Considering their lower cost, they offer better cost-effectiveness⁴.
• Application Resilience: Customers often use mixed-CPU architectures (combining Graviton and x86 instances) to improve overall application resilience. Testing application performance in a controlled environment before production deployment is recommended⁵.

Implementation Steps

1. Evaluate Workload Suitability: Assess whether your workload fits Graviton instances well. Consider factors like CPU requirements, memory needs, and I/O demands.
2. Transition Guide: Follow AWS’ transition guide to port your application to Graviton. Validate your workload’s performance in a test environment.
3. Cost Estimation: Use the AWS pricing calculator to estimate costs for running instances on different processors (x86 and Graviton). Compare costs and performance metrics.
4. Integration with Amazon EKS: If you’re using Amazon Elastic Kubernetes Service (Amazon EKS), integrate Graviton-based EC2 instances into your existing environment. You can mix Graviton and x86 instances to optimize cost and resiliency⁵.

Conclusion

Drop a query if you have any questions regarding AWS Graviton and we will get back to you quickly.

About CloudThat