Amazon's recently gone big with its burstable t2.large 8GB RAM instance and now the cloud giant has headed the other direction with a new t2.nano instance with RAM the size of the first $25 Raspberry Pi Model B -- and prices to match.
Amazon expects the t2.nano will be used to host low-traffic websites, run microservices, support test and development environments, and, like Raspberry Pi, it predicts it will be used in education and training.
The t2.nano can run 32- or 64-bit operating systems and applications and supports storage encryption and two Amazon's Elastic Network Interfaces per instance.
Amazon took the wraps off its the new t2.nano in October alongside its Internet of Things platform for building software to support IoT applications. The tiny instance offers one virtual CPU and 512 MB of memory, filling out the the company's line-up of services under its "burstable performance" model, which provides baseline performance that can be scaled up on call.
Other instances within this cost model include the t2.micro, t2.small, t2.medium, and most recently the t2.large. As always with AWS, it's all about the economics of compute power.
Baselines under t2 instances range from 10 percent on the t2.micro instance to 60 percent on the t2.large. Staying within the baseline means users won't chew through allocated CPU credits, which also grow over time. The availability of CPU credits in turn determines the duration an instance can burst for.
In the case of the t2.nano instance, the baseline is set at 5 percent and on-demand pricing costs $4.75 a month for the US, $5.11 a month in Europe, $7.30 a month in Asia Pacific, and $9.85 for South America. Alternatively a one-year reserved t2.nano instance in the US will cost $3.125 a month, equal to $37.50 per year. On a three-year reserved instance, the price drops to $25.20 for a full year.
AWS spokesperson Jeff Barr explains the CPU credit system for the t2.nano: "The t2.nano offers the full performance of a high frequency Intel CPU core if your workload utilizes less than 5 percent of the core on average over 24 hours. You get full access to the CPU core when needed, as long as you maintain a positive CPU credit balance."
"Each newly launched t2.nano starts out with a CPU credit balance of 30 credits, and earns 3 more credits per hour, up to a maximum of 72. This means that each instance can burst to full-core performance for up to 72 minutes at a stretch," he added.
Barr said the company had run numbers on how customers were handling its burst model credits and claims that during the course of couple of days over 96 percent of t2 instances were in the black on CPU credits.
"In effect, you are paying for a very modest amount of processing power, yet have access to far more when the need arises," said Barr.
The new instance isn't available yet in AWS's Frankfurt EU region or the Asia Pacific Sydney region but is available in all others including US East (Northern Virginia), US West (Oregon), US West (Northern California), Europe (Ireland), Asia Pacific (Singapore), Asia Pacific (Tokyo), South America (Brazil), and AWS GovCloud (US) regions.