Public cloud and enterprise datacenters continually require more computing power to meet the ever-increasing user demands. AWS processes billions of data requests every day1 from customers seeking performance, reliability and on-demand scalability in cloud instances that fit within their budgetary constraints. To meet its customers’ computing needs, AMD and AWS have collaborated to create distinct types of cloud instances designed to meet specific application needs: AMD-powered Amazon Elastic Compute Cloud (EC2) instances are available in four categories: general-purpose (M5a & M5ad), general-purpose burstable (T3a), memory optimized (R5a & R5ad), and now compute-optimized (C5a & C5ad).
Amazon EC2 C5a instances combine the power of the latest generation AMD EPYC processor with optional memory and storage configurations designed to support a wide variety of workloads such as data analytics, video encoding, gaming, image manipulation and more. With the broad range of instances available, the new C5a instances provide highly cost-effective cloud solutions with high performance, and the lowest cost per x86 vCPU in the Amazon EC2 family.
The C5ad instances extend the benefits of C5a with the ability to further tune workloads with low IO latency requirements using high-speed local storage caching, by adding local NVMe-based SSD block level storage connected directly to the host. C5ad instances come with up to 3.8 TB of NVMe based SSD storage and high-speed network connectivity. The high performance local NVMe storage and high-speed network connectivity in C5ad instances offer performance, value, and scalability to serve a variety of workloads.
Amazon EC2 instances powered by AMD EPYC processors are built on the AWS Nitro System—a collection of AWS-designed hardware and software innovations that enable the delivery of efficient, flexible, and secure cloud services with isolated multi-tenancy, private networking, and fast local storage—and deliver up to 10% cost savings over comparable instances3 in most regions, with the Asia Pacific (Mumbai) region offering up to 45% cost savings2, all while providing a reliable and scalable platform that brings optimal performance for enterprise-class workloads including web services and databases. Below are just a few examples of how AMD-powered Amazon EC2 instances are delivering predictable scaling and measurable impact.
Web and application servers for performance and scalability
Cloud-based solutions for distributed enterprise applications require a scalable infrastructure capable of accommodating dynamic capacity needs.
NGINX is a popular web server that can also be used as a reverse proxy, load balancer, mail proxy and HTTP cache. For high-performance, multi-threaded deployments, C5a can deliver significantly lower cost3 when implemented in the cloud. The following chart demonstrates performance of NGINX in both scale-up and scale-out deployment scenarios.
Read the NGINX on Amazon EC2 C5a Instance solution brief here.
High-performance, in-memory data store for real-time performance
Caching data and objects in memory can improve the throughput and often deliver near real-time data access performance.
Memcached is a popular, open-source, in-memory distributed caching system. There are several applications that can benefit from Memcached such as web application frontend, content delivery, media streaming, search engines, relational databases, gaming and many more. Amazon EC2 C5a instances powered by AMD EPYC processors are optimal in many ways for Memcached, offering cost effective, high performance and scalability on demand. Memcached on Amazon EC2 C5a instances can offer predictable performance starting with the application’s current needs and as requirements grow. Our benchmark tests demonstrate the performance and scaling for both scale-up and scale-out scenarios as demonstrated below.
Read the Memcached on Amazon EC2 C5a Instance solution brief here.
In addition, I wanted to highlight the recent performance characterization of Redis Enterprise on AWS C5a instances in collaboration with Redis Enterprise from Redis Labs, a real-time database and enterprise grade caching layer.
Powerful performance in business transactions and decision support
Relational Database Management Systems remains the core of enterprise applications for transaction processing, business analytics, and decision support systems.
MySQL is one of the most popular open-source relational database management systems. Implementing MySQL in the cloud is an increasingly popular choice for many applications. Performance, scalability, security features, reliability, and cost of ownership are all important factors when choosing a platform for a MySQL deployment – Amazon EC2 C5a instances offers all of them. We have tested a well known Online Transaction Processing (OLTP) benchmark on AWS EC2 C5a instances and a Decision Support System (DSS) benchmark on AWS EC2 C5ad instances which offer high speed local storage for tempdb to analyze the performance and scalability of MySQL. These results, shown below, demonstrate the effectiveness of AWS EC2 C5a instances in common relational database deployment scenarios.
Read the MySQL on Amazon EC2 C5a Instance solution brief here.
Big Data applications for deeper insights
Enterprises across industry verticals are realizing the power of Big Data Analytics for gaining operational efficiency for new business opportunities.
Apache Hadoop offers an ecosystem of open source components that fundamentally changes the way enterprises store, process, and analyze data. Cloudera Distribution Hadoop (CDH) is the most popular distribution of Hadoop. The new C5ad instance is an optimal fit for Apache Hadoop based workloads. Combining the high performance of AMD EPYC with high performance local NVMe storage for temporary storage with the right network bandwidth can match demanding workload requirements and achieve predictable scaling of performance as shown below.
Fast Encoding/Transcoding for live streaming
We are in a new era of high-quality video in social, entertainment and business applications. Such applications in many cases, use real-time encoding/transcoding using FFmpeg like frameworks. FFmpeg is the leading open source multimedia framework used to decode, encode, transcode, mux, demux, stream, filter, and play videos. The benchmark testing of encoding time for the Tears of Steel movie clip, shows real-time video delivery performance by AMD EPYC powered Amazon EC2 C5ad.8xlarge (32 vCPUs) instances. The high core count and exceptional memory bandwidth of the AMD EPYC processor combined with high-speed NVMe local storage on C5ad instances, enable fast encoding/transcoding. See chart below showing faster encoding time for 1080p and 4K using NVME local storage showing real-time video stream delivery! This performance testing used the VP9 codec in Constant Bitrate mode with the Highest quality setting for streaming.
In short, the latest generation of AMD’s record-setting EPYC processors power the Amazon EC2 C5a and C5ad instances, giving customers a variety of options for high performance, scalability, reliability, and security features -available on-demand with pay-per-use pricing at the lowest cost per x86 vCPU in the Amazon EC2 family.
For more AWS Cloud solutions powered by AMD EPYC processors, visit AMD EPYC Tech Docs and White Papers Library
- “The AMD-based instances provide additional options for customers who are looking to achieve a 45% cost savings on their Amazon EC2 compute environment for a variety of workloads.” See https://aws.amazon.com/about-aws/whats-new/2019/11/amazon-ec2-amd-instances-are-now-available-in-asia-pacific-mumbai-aws-region/
- AMD powered AWS EC2 instances offer up to 10% lower cost compared to comparable x86 based instances. See https://aws.amazon.com/ec2/amd/
Raghu Nambiar is a Corporate Vice President for AMD. His postings are his own opinions and may not represent AMD’s positions, strategies or opinions. Links to third party sites are provided for convenience and unless explicitly stated, AMD is not responsible for the contents of such linked sites and no endorsement is implied.