In the hunt for more cloud customers, Microsoft today announced it has cut in half the price of its entry-level server for its Azure infrastructure as a service. Instead of $.04 an hour, it will be $.02 an hour for the Windows server, said Steve Martin, Windows Azure general manager.
Microsoft is responding to recent Amazon Web Services EC2 cloud price cuts, announced earlier this week. Amazon charges $.02 for its entry-level, Linux micro server, unchanged by this week's price announcements. On Monday, Amazon dropped the price of its next-in-line, small-server instances, which run either Linux or Windows. Amazon's small Linux instance dropped from $.085 to $.08, and a small Windows instance from $.12 to $.115.
[ Want to learn more about the pluses and pains of cloud computing? See Four Cloud Computing Pain Points That Still Hurt. ]
Until now, Microsoft had not sought to match Amazon's micro Linux server price with its own entry-level Windows server. Windows typically commands a premium over Linux in the cloud.
Microsoft's next size up in its Azure server offerings is its small server at $.12 an hour, unchanged from its previous listing.
Hidden behind the pricing tables is a difference in the virtual CPU power provided by each small and micro instance. Amazon defines an elastic cloud compute unit (ECU) as the equivalent of a 1- to 1.2-Ghz Xeon core available in 2007. Microsoft's Martin defined an Azure virtual CPU as the equivalent of a 1.6-GHz Xeon.
"An hour of processing in Azure is going to yield favorable benchmark results over an hour in Amazon," Martin asserted.
There are other differences between the two service providers' lower-level offerings. Amazon defines a micro server as having 613MB RAM and access to the Elastic Block Store service. Microsoft defines it as having 768MB RAM and 20GB local disk storage. Small instances from Amazon come with 1.7GB RAM and 160GB of local disk; from Microsoft, 1.75GB RAM and 225GB local storage. In other words, Amazon is setting a marketplace standard on demand-compute instances and Microsoft either matches it in each category or offers something a little larger.
Microsoft also reduced its on-demand storage pricing 12%, from $.14 to $.125 a gigabyte per month. Six-month storage plans were reduced up to 14%. Microsoft and Amazon now have the same per-gigabyte charges for the first 49 gigabytes of long-term storage.
Martin said the most frequent comparison between Microsoft and Amazon uses the price list from Amazon's heavily trafficked Northern Virginia data center or U.S. East -1. In fact, prices vary between Amazon's 7 regional centers, with U.S. East and U.S. West in Oregon having the lowest. "We have globally consistent pricing," said Martin.
Microsoft maintains cloud data centers in Chicago, San Antonio, Singapore, Dublin, Ireland, and other locations.
Martin also claimed Microsoft is maintaining a simpler price list, with fewer options skewed toward large, regular customers. "We're going the other way. We want to make our price list as simple as possible. We don't want customers burning mental cycles on those problems," he said.
Microsoft reduced pricing on its Azure-based database system or SQL Azure on Feb. 14. It created the option of using a 100MB database at $5 a month, or half the previous price. Customers will see savings of 48% to 75% on database sizes larger than a gigabyte.
Google's App Engine cloud reduced storage prices earlier this week as well.
The series of price reductions in the cloud appear to have followed on the heels of Amazon's VP of product management Adam Selipsky's remark that Amazon was going to cut more than just the price of storage in the cloud in the future. Price cutting is part of its "retail DNA," he said in an interview.
All cloud suppliers are seeing the benefits of strong CPUs in the servers they buy and greater storage capacities in disks, giving them elbow room to lower prices. The cloud service provider market is likely to see periodic rounds of small price cuts as cloud data centers replace old equipment with new.
Find out how to move beyond server virtualization to build a more flexible, efficient data center in the new Private Cloud Blueprint issue of Network Computing. (Free registration required.)