IT pros welcomed the addition of a new AWS T2 instance size to the burstable line of Amazon machine images, but...
some say the instances aren't suitable for use with production applications.
The Amazon Web Services (AWS) T2 line of instances now has a T2.large size available, with 8 gigabytes (GB) of RAM, doubling the amount of memory previously available with T2.medium instances. While the T2.large still has two virtual CPUs (vCPUs), its baseline performance is set at 60% of those CPU cores and it accrues 36 CPU credits per hour, which can later be exchanged for a burst in performance over baseline.
"Our customers told us that the burst-based model gave them plenty of CPU power to run applications that consumed large amounts of memory," wrote Amazon chief evangelist Jeff Barr in a blog post. "The new size provides double the amount of memory, along with a higher baseline level of CPU power."
AWS T2 instances are good low cost alternatives for mostly idle environments that are required to be available 24/7/365, such as test and development, said Hartman Wagner III, principal software engineer for Progress Software Corp of Bedford, Mass.
The larger memory sizes also mean customers can create low-cost test environments that more closely match production, but at a 25% to 50% savings, Wagner said. However, they are not for use with production applications.
"If you shut down or reboot the instance, you lose your performance credits and the credit window is only 24 hours," Wagner added. "So you can't build up for days and then get a full powered system for a day."
Some cloud consultants have mixed feelings about the AWS T2 line. While less expensive per hour than comparable M series instances, they max out at 60% CPU power with the T2.large, so customers might pay more for the same number of CPU cycles if they don't schedule the T2 line correctly, according to Kris Bliesner, CTO of 2nd Watch, Inc., an Amazon Premier Partner in Liberty Lake, Wash.
"The T2 line is appropriate if you don't have to guarantee performance," Bliesner said. "But it's not appropriate for production apps in the enterprise."
Still, some IT pros want Amazon to expand the T2 line.
"The new T2.large is a great step in the right direction," said Brian Tarbox, a lead engineer working for a stealth startup in the Northeast. "Eight gigabytes of memory is still less than what a lot of applications need, so I hope we someday see a T2.xlarge."
The burstable instances start with an initial bucket of CPU credits, so if an app needs its burst at startup, rather than after an extended quiet time, they can take advantage of the burstable instance types, Tarbox said.
Meanwhile, the warnings about production use haven't stopped some enterprises, such as the Boston Celtics, from using AWS T2 instances in production, for eight servers whose workloads are a perfect fit for the T2's burstable performance pattern.
"The new t2.large instances could be a great option for small and midsize businesses already running workplace applications on m3.large or other legacy large instances," said Andrew Dawson, solutions architect at Boston-based G2 Technology Group, an Amazon partner that worked with the Celtics on deploying T2 instances. "We have had nothing but success with less intensive workplace applications on the t2.mediums."
Switching over to the t2.large instance size can save a customer almost $1,000 a year with on-demand pricing, which is appealing for SMBs who are trying to run their infrastructure as efficiently as possible, Dawson said.
Other AWS customers run development environments, small databases, application servers, and Web servers on their T2 instances, according to Barr's blog post.