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.

The Uberization Of The Data Center

Documents in a San Francisco taxicab bankruptcy filing a few weeks ago blamed Uber in part for its downfall. The filing was emblematic of the “Uberization” taking place in many sectors that are being digitally disrupted, including enterprise IT. If you’re the business being Uberized, the term is pejorative for how your big, expensive sales wins have been replaced by far smaller drips of as-a-service spending, only when and where customers need it, like Uber, Lyft and Zipcar.  

On the other hand, if you are the customer, Uberization is about streamlined procurement, new levels of immediate delivery, and the ability to pick and choose even the most arcane features you need from the swipe of your smartphone or a click of a Web browser.

One might consider Amazon Web Services as the first Uberizer of the data center world. Its revenue ramp from $0 in 2006 to almost $8 billion business 10 years later --  providing 10x more computing capacity than the next 14 cloud competitors combined -- defines the term “transformational.”

What Amazon did was to allow customers to divide resources into increasingly thin slices and do unheard-of procurement practices – buying exactly what they want - from a simple Web browser:

  • Order only what you need
  • Order only when you need it
  • Pay per usage (or IOPs or GB)
  • Order something different at any time
  • Lower total cost to below that of a purchase; unheard of either with Uber or old-style leases.  

Like Uber, AWS changed not only in how IT is procured, but how it is consumed.  And the trend is happening in sectors beyond computing resources. Among IDC’s FutureScape top 10 predictions for enterprise infrastructure in 2016, was the prognostication that by 2020, 80% of IT infrastructure will be bought on a pay-as-you-go basis.

We’re already seeing the same changes in the networking and storage domains. For networking, virtualized mobile networking as a service (NaaS) is allowing data center managers to connect together multiple WANs, even when they are continents apart and operated by different companies.

For storage, IT Brand Pulse examined AWS’ stellar 3Q 2015 results in depth, including its cloud storage revenue, and predicted that by 2020, 50% of all storage will be provided on an OpEx, pay as you consume, as a service basis. By 2023, that figure jumps to 80%.

So what does Uberization mean for data center managers?

  1. Lower data center costs. UC Berkeley Reliable Adaptive Distributed Systems Laboratory (RAD Lab) predicts a 75-80% reduction via a combination of greater purchasing power, management efficiency and a laser-focus on lowering costs.
  2. Agility, at last. Elastic resources that scale up and down, and enable businesses to be flexible with market conditions and customer needs, will offer a way to lower costs and gain a competitive edge.
  3. Ease of use that saves time. For decades, IT managers have endured 6- to 12-month long purchasing cycles, the need rip-and-replace 5-year old infrastructures, and very messy management interfaces. No more.
  4. Job growth.  US Bureau of Labor and Statistics data from December 2015 actually predict a 17% increase in software developer jobs and a 13% increase in IT jobs in the next 10 years, although computer programmer jobs will decline by 8% as they move offshore. Anecdotal replies by IT managers on a Quora thread on whether cloud computing will kill IT jobs reflected a similar trend. “Yes, (cloud computing will impact IT jobs), if you are making a living by clicking NEXT in OS or software installations. But no, if you are solving real-world problems, like how to script actions and make sure they work perfectly 99% of the time," wrote one IT manager.

Like any transition, the Uberization of data centers will never be an all-or-nothing event. Security concerns will always necessitate isolation of certain assets behind a firewall. Latency or bandwidth requirements may suggest a public cloud isn’t the best way to run every application. Architectures still will need to be built with forethought so they perform to SLAs.

But at the end of the day, people running data centers are the same as people who have chosen newer ride-sharing options over the San Francisco taxicab company. When the market makes it easy for them to buy exactly what they want, they vote with their wallets.