Laptop Displaying the GigaOm Research Portal

Get your Free GigaOm account today.

Access complimentary GigaOm content by signing up for a FREE GigaOm account today — or upgrade to premium for full access to the GigaOm research catalog. Join now and uncover what you’ve been missing!

How the mobile-first world will transform the data center

Table of Contents

  1. Summary
  2. Introduction
  3. The challenge of fixed-line web infrastructure
  4. The mobile-revenue shortfall
  5. The different workloads of mobile
  6. Tomorrow’s much leaner data center
  7. Software for low power: slim web clients
  8. Toward tomorrow’s lean data centers
  9. Recommendations
  10. About Martin Piszczalski

1. Summary

Growing at 40 percent annually, smartphone shipments are overtaking those of personal computers. Various forecasts predict that in four years there will be 1 to 2.6 billion smartphones, which will far outnumber PC shipments. But per-user infrastructure costs are not moving in concert with this dramatic revenue drop. In particular the back-end data center continues to operate as if it is still the halcyon days of ever-growing ad revenues.

Look for a new generation of mobile-centric data centers to arise over the next three years, with chips, servers, and power architectures customized for mobile workloads.

Key highlights from this report include:

  • The mobile market is undergoing tremendous growth, and rapidly. Data center infrastructure, on the other hand, has been slow to catch up, due to the longer time frame required for architectural changes.
  • Mobile advertising does not generate revenues that support the number of users, which costs some web companies billions of dollars.
  • Tomorrow’s data centers must allow savings on power. Data center operators can save by reserving power from the utility, making better cooling decisions, and using servers conservatively.
  • The “mobile first” philosophy is under way today. In other words, the industry is beginning to design directly for mobile. Such a change could reduce the back-end workload of an application.