First impression on unpacking the Q702 test unit was the solid feel and clean, minimalist styling.
Amazon moves to next-generation compute instances
- — 21 January, 2014 20:29
Sensing the customer need for more computational muscle in smaller packages, Amazon Web Services has added a number of new packages in its Elastic Cloud Compute (EC2) service.
The Amazon subsidiary also reduced prices on selected storage packages, reducing the cost of its Simple Storage Service (S3) by up to 22 percent and Elastic Block Service (EBS) by up to 50 percent.
The new packages, or compute instances, in the EC2 line fill out the company's second-generation M3 line of instances, which the company is now touting as the successor to the original M1 line. Amazon Web Services (AWS) introduced the M3 series a year ago as a more performance-minded alternative to the base M1 offerings.
The M3 series features faster CPUs, making them better suited for compute-intensive tasks such as video encoding, batch processing and caching. The M3 line also features SSDs (solid-state disks) which, according to AWS, offer a more consistent I/O rate than M1 services.
Previously, customers could only get the M series instances in four or eight virtual CPU packages. The two additions to the M3 family provide smaller instance sizes, with one or two virtual CPUs. The M3 instances run on Intel Xeon E5-2670 (Sandy Bridge or Ivy Bridge) processors.
AWS is pitching the M3 instances (called M3.medium and M3.large) for general computing tasks -- such as maintaining a small or medium-sized database, or for running an enterprise application -- that the customer might have run on M1 instances before. AWS claims that, for these workloads, the M3 instances provide more consistent performance and are more cost-effective.
M3.medium offers a single virtual CPU, 4GB of solid-state storage space and 3.75 gibibytes of memory (GiB), and costs US$0.113 per hour. By contrast, M1.medium offers a single virtual CPU, 3.75 GiB of memory and 160GB of disk storage, for $0.175 per hour.
A gibibyte is slightly larger than a gigabyte, because it is a binary multiple of a byte. For instance, a standard megabyte offers 1,000,000 bytes whereas as a mebibyte offers 1,048,576 bytes.
M3.large offers two virtual processors, 32GB of storage and 7 GiB of memory, at the cost of $0.225 per hour.
The price cuts on AWS' other services are not the first for the company. AWS has made 40 price cuts in various services since 2006.
In the S3 line of cloud storage services, prices have been dropped for both the standard storage pricing and the reduced redundancy storage. For up to the first terabyte, AWS now offers storage at US$0.085 per GB per month, which is an 11 percent reduction.
Greater cuts were made for larger pools of storage. Customers storing more than 5,000 terabytes will now be charged $0.043 per GB per month, a reduction of 22 percent.
Prices on EBS have also been cut, for both the amount of storage and the I/O requests. EBS is AWS' preferred method of storing data for EC2 compute instances.
The price cuts for EBS vary by region. For the U.S. East Coast region, a GB of provisioned storage per month will run $0.05, a 50 percent reduction from the previous price of $0.10 per GB per month. The European Union region will also see prices halved: A GB of provisioned storage per month will run $0.055, a reduction from $0.11 per GB per month. For the Asia Pacific region covering Tokyo, a GB of provisioned storage per month will now cost $0.085 per GB per month, down from $0.12 per GB per month.
The cost of EBS I/O (input/output) requests has also been reduced. Now, a million I/O requests will cost $0.05 rather than $0.10, a 50 percent reduction in cost.
AWS routinely cuts the prices of its cloud services in what Andy Jassy, its vice president for cloud services, has called a virtuous cycle, where the more customers AWS signs up, the better economies of scale it can enjoy. Because it can buy more equipment, AWS can negotiate for steeper discounts with suppliers. It can also pioneer ways of standardizing and automating how it manages its infrastructure, further cutting the cost of maintenance.
The new pricing tiers will take effect on Feb. 1, 2014.