Advanced Computing in the Age of AI | Friday, March 29, 2024

Amazon T2 Cloud Instances With Burst Capability 

Amazon Web Services is shifting down market with a new Amazon Elastic Computer Cloud (EC2) instance type designed for web and database applications that benefit from a performance burst but don't require sustained high CPU horsepower.

AWS said pricing for its on-demand T2 instances start at $9.50 per month (or $0.013 per hour), making them the cheapest Amazon EC2 option. The cloud provider is targeting the general-purpose instance type for developer environments, small databases and web servers.

"T2 instances address this need by providing a consistent baseline performance with the ability to burst to full CPU core performance," Matt Garman, AWS vice president for Amazon EC2, said in a statement.

Still, at least one initial assessment of the new T2 instances warned that potential cost savings might be offset by the need for additional storage.

Amazon's EC2 service is designed to provide adjustable computing capacity in the cloud, allowing users to scale applications up or down depending on requirements. It is also billed as reducing the amount of time required to obtain and boot new server instances.

AWS said in a blog post the T2 instances are based on a "processing allocation model" that provides a baseline amount of processing power combined with the ability to scale up to a full core when needed. The "burst" capability is based on the number of "CPU credits" customers accumulate during quiet periods and how many are spent while running heavy workloads.

Customers "can provision an instance of modest size and cost and still have more than adequate compute power in reserve to handle peak demands for compute power," AWS claimed.

T2 is available in three instance sizes (micro, small, and medium) from datacenters in the U.S. East and West regions along with facilities in Ireland, Singapore, Tokyo, Sydney, Australia, and Sao Paulo.

AWS also released performance specs for on-demand T2 instances from its Northern Virginia datacenter. The low-end t2.micro priced at $0.013 per hour includes 10 percent of single-core performance from a physical CPU allocated to the instance. The T2 instances are based on a single core of an Intel Xeon processor running at 2.5 GHz. The micro option also includes 1 GB of memory. The t2.small option provides 20 percent of Xeon single-core performance and 2 GB of memory for $0.026 an hour ($19 per month). A t2.medium option includes 40 percent of single-core performance and 4 GB of memory for $0.052 ($38 per month), AWS said.

Each T2 slice gains "CPU credits" at a rate determined by instance size. Each credit provides full CPU core performance for up to one minute.

For example, a micro instance receives six credits per hour. The total number of hourly credits doubles for the each of the other options. The credits are "spent" when the instance is active, and can be stored for up to 24 hours.

The new instances are being viewed as a likely replacement for previous generation M1 instances. AWS noted that the T2 instances do not include local instance storage and that at least one Amazon Elastic Block Store volume would be needed for local storage.

"It’s important to note that although the new t2.small instances are 40 percent cheaper than the m1.small instances, the m1.small instances include storage, while the new T2 instances do not," the Rightscale cloud management blog pointed out.

Among the uses cases cited by AWS for its new instance type was a cloud application management platform from Engine Yard. The platform-as-a-service runs "industrial strength applications" for ecommerce and other vendors.

While the prototyping and staging environments for deploying production applications typically run 24/7, they can be underutilized when not handling a production load request, the company noted. The new T2 instances could be run on the Engine Yard platform to reduce infrastructure costs without sacrificing performance, claimed Engine Yard CTO Rob Walters.

AWS said T2 instances will be launched "in the near future" in northern California, Beijing and as part of AWS government cloud services.

About the author: George Leopold

George Leopold has written about science and technology for more than 30 years, focusing on electronics and aerospace technology. He previously served as executive editor of Electronic Engineering Times. Leopold is the author of "Calculated Risk: The Supersonic Life and Times of Gus Grissom" (Purdue University Press, 2016).

EnterpriseAI