/>
X
Innovation

Google: The cloud is cheaper

Following a recently published report from McKinsey which cast a shadow on cloud computing, Google says to consider the whole picture, not just hardware costs.
victoriaho.jpg
Written by Victoria Ho, Contributor on
When considering the costs of going to the cloud, consider the whole picture, not just hardware costs, said a Google executive.

Following a recently published report from McKinsey (below), which cast a shadow on cloud computing, the cloud platform vendor is perturbed.

The search giant spoke to ZDNet Asia Thursday in response to the report, which recommended large companies be better off managing their IT resources in-house. It compared the costs of running a typical enterprise data center against a cloud alternative, and found the latter to be a more costly option for large enterprises.

Matthew Glotzbach, product management director, enterprise, at Google said the report was flawed in that it only focused on hardware cost.

"It is only one cost aspect out of a dozen." Companies also need to consider other costs of the application stack such as licensing and development, said Glotzbach.

Furthermore, he said the report also assumes virtualization will be able to raise hardware utilization to 100 percent. Glotzbach said, in reality most companies are only able to achieve 20 to 30 percent.

"Enterprises haven't been sophisticated enough to plan for that high utilization level." Virtualization, to achieve high utilization "is good in theory, but in reality it isn't even close to how enterprises are run", he said.

Google has been on the path to win large clients over with its cloud-based application offering, Google Apps.

Glotzbach said the majority of its 3,000 new customer wins each day have been small and midsize businesses (SMB), but pointed to the "few dozen large enterprises adopting [cloud services] every quarter" as an indication of cloud services maturing.

He added that the cloud's value lies beyond cost reduction--as a platform, it can help serve functions that companies could not otherwise achieve in-house.

Within the Google Apps offering, he said, companies can post videos for streaming. "There's no way companies could have set that function up themselves," said Glotzbach.

On the cloud, companies can adopt the latest technology without the investment or development cycle involved with building apps in-house, he added.

Companies should think about moving commodity aspects of the business such as e-mail and calendaring to the cloud, he said. Core applications central to the business should be kept in-house; companies need to decide if they want to rewrite these applications to the cloud.

Glotzbach noted that industry momentum around cloud platform providers making it easier for developers to port existing apps to the cloud will help bridge this divide in future.

What McKinsey said
McKinsey released a report in April 2009, which said cloud computing is suited for smaller organizations, not larger corporations.

In a study, the consulting group found that for small to midsize businesses (SMB), the cost of running a data center is higher than the cost of transiting to a cloud computing environment. Large enterprises, on the other hand, should run on their own data centers, McKinsey recommended, saying it is much cheaper to manage in-house compute capacity.

Using Amazon's EC2 cloud platform, as an example, the report revealed an increase of 144 percent in costs is needed to move to the platform.

Furthermore, non-labor costs increased from US$43 per month for typical data centers to US$270 per month for Amazon's cloud. The report also said companies can match server utilization rates in the cloud by applying virtualization technology in-house. It said companies can achieve server utilization rates of about 35 percent with "aggressive virtualization" and a "generally achievable" rate of 25 percent. Going to a "best-in-class" option, using Google as an example, would yield about 38 percent.

Large companies should instead rely on server virtualization to share compute resources in a more cost-effective way, according to the report.

Konrad Foo contributed to this story. Based in Singapore, he is an intern with ZDNet Asia.

This article was orignally posted on ZDNet Asia.

Editorial standards

Related

How to use your phone to diagnose your car's 'check engine' light
BlueDriver Bluetooth dongle

How to use your phone to diagnose your car's 'check engine' light

Hackers are finding ways around multi-factor authentication. Here's what to watch for
a-man-looking-at-his-smartphone-while-sitting-at-a-computer-in-his-home-office

Hackers are finding ways around multi-factor authentication. Here's what to watch for

Elon Musk drops details about Tesla's humanoid robot
tesla-humanoid

Elon Musk drops details about Tesla's humanoid robot