Network Computing is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Data Center Decision Time: Stay Or Go?

 

 
InformationWeek Green -  Mar. 4, 2013 InformationWeek Green
Download the entire June 10, 2013, issue of InformationWeek, distributed in an all-digital format (registration required).
 

 

 

Data Center Decision Time

Bits and bytes are the basics of our business, and business is good. So much so that even the immense capacity of advanced data centers can't keep up with demand. And complexity? Off the charts. Now the question is, does continuing to manage your own data center facilities make financial sense?

That might seem like a contrarian takeaway from our InformationWeek 2013 State of the Data Center Survey, given that all respondents are involved with management or decision-making at organizations with data centers of 1,000 square feet or larger. In fact, the percentage of respondents whose operating centers are at least 25,000 square feet jumped four points from our 2012 survey, to 15%.

The enterprise-owned glass house is alive and well: Our survey shows just 4% of respondents aggressively moving applications to the public cloud, and another 4% of survey respondents expect demand for data center resources to decrease, compared with 69% expecting increases. Many are girding for upticks of 25% or more. At that pace, enterprises can't further virtualize or build new facilities fast enough to keep up, even if they could afford to -- which many can't, given that spending on data center facilities and associated infrastructure is little changed year over year.

After you've consolidated existing facilities and servers to the limit, what then?

To get more done with the same physical resources and people, CIOs might look to such cutting-edge technologies as self-service private clouds; software-defined networks and storage; and high-density, low-power servers. They can off-load commodity IT services such as email, unstructured data backup and public-facing websites to public cloud and software-as-a-service providers. But is that enough, given how fast demand for application resources, storage, network capacity and services is rising?

In the column on p. 6, we discuss the U.S. government's massive consolidation effort in which it will close about 40% of its data centers and make a lot more use of the cloud. It's a good start, but it's not a full-on move away from current models. Even if the Office of Management and Budget's goal is met, Uncle Sam will still maintain almost 2,000 data centers. In the private sector, FedEx is moving to a hybrid private-public data center model. Its new Colorado Springs facility uses a private cloud architecture, whereby any company workload can run on commodity hardware. FedEx plans to meet some of its future capacity needs by moving certain Tier 1 applications to the public cloud, to avoid having to invest more in its own data centers.

But few CIOs of established companies have charted such a clear strategy. Our take is that they have a bad case of the paradox of choice: Faced with myriad alternatives -- private data centers, leased wholesale space, collocated facilities, managed infrastructure services, virtual private clouds, public IaaS, PaaS, SaaS -- they get overwhelmed and become paralyzed.

Report Cover

Our report on the state of the data center is free with registration. This report includes 38 pages of action-oriented analysis, packed with 30 charts.

What you'll find:

  • Top 14 success metrics for server virtualization
  • Breakdown of total IT budget spent on data center ops

Get This And All Our Reports

To get unstuck, remember the decision shouldn't be: Do I build, upgrade or lease data center infrastructure? Rather, it's: What's the optimal way to deliver this particular application or IT service? The answer should be based on analysis of requirements, including availability, reliability, security, redundancy, flexibility to meet business and application requirements, compatibility with existing infrastructure and, of course, cost (however you measure that: TCO, opex, capex).

Time is running out. Only 55% of respondents to our survey say they can continue to add servers without needing to build new facilities -- that's a 10-point drop from last year. Meanwhile, the percentage of respondents planning to move IT-owned systems to colocation facilities rose by just five points, to 20%. We hope the benefits of leased space that we highlighted in last year's report are becoming apparent to IT decision-makers, but we're early in this transition. Less than 40% of respondents put one-tenth or more of their existing systems in third-party facilities, though the percentage jumps six points, to 45%, for new systems. Given the expense and complexity of building and operating private data centers, those percentages are lower than we expected. On the plus side, public cloud services are a more popular option for new applications, particularly among startups and small enterprises without much (if any) legacy infrastructure. More than half of survey respondents say 10% or more of their new applications will use, in whole or part, multitenant IaaS or SaaS services instead of a dedicated private infrastructure. For an intrepid 8% of respondents, the cloud will be the platform for half or more of their new applications -- again, a percentage that's lower than we expected. At the far end of the spectrum are companies like Netflix, which has more than 90% of its revenue directly dependent on a cloud-hosted streaming service.

To read the rest of the article,
download the June 10, 2013, issue of InformationWeek.