The Seamless Enterprise

Comprehensive news and discussion of enterprise communications and converged network solutions.

500ms Is Too Much Time

on December 20, 2011 by Christopher Glenn

Over the next decade, the focus of network evaluation will shift toward latency. In the client- server model, connectivity was in the background. In a cloud-based model, connectivity is front and center. This is because a cloud-based model relies on network connectivity as each letter is typed.

You have probably seen this without realizing it because web pages today use a lot of Ajax, a language that makes a simple web form behave like a cloud application. For example, when doing a Google search with auto-complete turned on, your search results are updated as every letter or word is added to your query. In another example, a section of a web-based form changes based on your selection of a radio button (e.g., if you change a button on a form from “individual” to “company,“ a new field might magically insert itself into the form without the whole page being reloaded.)

Any delay over 500 ms (milliseconds) is likely to be noticed by users. One source of delay is the bottlenecks in the device or local system (local delays can be caused by too much multitasking in the microprocessor, not enough RAM or hard disk page memory, and issues with hard drive utilization.) The other primary source of delay is the network, typically from the devices that route network traffic and  serve up content in response to a network request. In other words, the biggest components of “network” delay are rarely the network itself.

Regardless of the source of this delay, the cloud computing era will drive a laser-like focus on delay and network latency. If you have ever tried to mute your smartphone or Unified Communications client only to experience a three- to five-second delay as you watch an “hourglass” or “spinning wheel of death” before your phone mutes, you know what I am talking about.

What will slow the adoption of cloud technologies is the lack of tools to help users understand the source of delay. Without this, end users will blame “the network” and that does little to help mitigate the problem. I have seen entire cloud-based application deployments fail because companies didn’t invest enough in cloud-based server resources.  This is a shame, because a simple time and motion study can tell you that the difference between 200 ms and 1200 ms is a lot more than one second. The difference is success or failure. 

A simple time and motion study would help quantify the difference. If 500 people use an application for 10 minutes per day, and a typical user has to wait an extra second (1000 ms) at a rate of three times per minute, that’s 250 minutes per day or close to 100 hours per month. If the average user’s wages are just $25 an hour, that’s close to $2,500 per month in lost productivity due just to the delay. However, when you include the lack of compliance and failure of entire initiatives that result from people being fed up with the application’s performance, that same delay could cost your enterprise millions. Certainly, it would be worth an extra $500/month in infrastructure to mitigate the problem.

The biggest problem in technology investment today is cutting corners on investment. If you invest in a cloud-based server infrastructure with the rationale that you will add more capacity when it proves to be successful, you might be causing the investment to fail. The difference between one second and five seconds to update a cloud-based app is likely the difference between success and failure. It’s unfortunate that basic operations paradigms have failed to find their way into IT to help ensure the correct level of infrastructure investment.


Comments (0) Leave a Comment

Add a comment:

Name:
Email:
Website:

  • Comment
  • Preview
Loading


About the Author

Christopher Glenn explores emerging technologies to help companies create convergence strategies that bring together wireless and wireline communications. He has 25 years of experience in the telecommunications industry, with roles spanning strategic planning, business development, operations, engineering, sales, marketing, and finance. Christopher's career includes over 10 years with Sprint, most recently as General Manager of Converged Business Solutions, where he focused on the company's managed services portfolio, VoIP and IP telephony and mobile integration. He holds a BSB with distinction in general management and finance as well as an MBA with honors in corporate strategy and operations management from the University of Minnesota's Carlson School of Management. Follow him on Twitter at http://twitter.com/NetThink.

Share

More news
from sprint

Register here to receive
future newletters
from Sprint.

Register