Amazon Web Services, fresh off an outage that brought down big-name sites such as Reddit and Imgur, today announced an 18% price reduction for its virtual machines, the 21st time the leading infrastructure as a service (IaaS) vendor has dropped prices since launching its cloud in 2006.
In addition to the price drop, AWS released a new series of Elastic Cloud Compute instances with high input/output (I/O) qualities. They're optimised, AWS says, for media encoding, batch processing, caching and Web serving. The extra-large instance (m3.xlarge) comes with 15GB of memory and 13 ECU -- which are Amazon compute units -- across four virtual cores. A double extra-large instance has 30GB of memory with 26 ECUs on eight virtual cores. The service debuted in the Northern Virginia US-East region, but AWS plans to roll it out to other regions early next year.
The new prices are as follows:
" Small instances went from $0.08/hour to $0.065/hour " Medium instances went from $0.16/hour to $0.13/hour " Large instances went from $0.32/hour to $0.26/hour " Extra-large instances went from $0.64/hour to $0.52/hour
Dan Feld, who manages sales and business development at AWS consultancy Newvem and is a former AWS exec with the same role for Amazon in Europe, says the price drop for the small instances is a significant move for a lot of customers. According to a sample of 40,000 instances that Newvem monitors for customers, m1.small is the most popular type, making up more than 25% of instances. All of those customers will have a drop in prices. Along with the price drop, AWS is filling out its product portfolio with higher-end instance types as well, he says.
"Amazon sure doesn't want to make it easy for competitors," says cloud analyst Paul Burns of Neovise about the company's 21st price drop in six years. As processing performance continues to improve, Burns says AWS has taken the approach that it will pass those savings on to customers whenever it can, as opposed to waiting for some major breakthrough in performance or cost.
It's the same idea on the new instances. The high-end instance types follow a trend by AWS in recent years. AWS has rolled out high-performance compute options, cluster compute instance types, and now these new large and extra-large high-memory offerings.
AWS is also playing defense here, though. Burns says as the cloud market continues to evolve, more and more vendors will come out with offerings aimed at biting into AWS's market share. While he doesn't see any connection with this announcement and the recent outage, he does see more customers looking outside of the AWS ecosystem to providers who may serve a niche need, or perhaps to another cloud provider as a backup. "Amazon's still in great shape," Burns says, but other competitors seem to be champing at the bit.
The news comes almost two weeks after Amazon experienced its third major outage in two years. On Oct. 22, a memory bug caused by new hardware that had been installed in a Northern Virginia data centre caused a cascading effect that eventually brought down a large portion of one of AWS's availability zones.