Performance matters

By: Brandon Harvey

Performance matters.  For people who work closely with technology, this is something of a ground truth.  Faster is better; it has always been thus.

But how much does performance matter?  Can we quantify its value, so it can be weighed rationally against other attributes a system might have?  Generally speaking, this is difficult, but in some ways it’s getting surprisingly easy – particularly, in the context of the Web.

Here’s how it goes: take a website with a high volume of traffic.  Change some attribute of the website, in some small way, and expose a certain fraction of the audience to that alternate version.  See what happens.

This is A/B testing.  It’s not a new idea, and major websites have used it for years to help with layout and design tuning.  But recently, some companies have played with A/B testing in another way: to test user’s reactions to different levels of performance.  And the results, while not surprising, have been fascinating.

Both Google and Bing, for instance, ran studies where they deliberately returned search results more slowly, for a small subset of (unsuspecting) users.  Over a series of weeks, they kept exposing the same users to a system that was slightly slower than the norm.

And as it turned out, the users of the slightly slower search engine . . . searched less.  Some portion of searches that would otherwise have happened, just didn’t happen.  Users got less utility out of the system.

And it only took a small slowdown to create this effect.  As little as 100 ms of delay (a tenth of a second) had a measurable impact on users.  This is subtle stuff.  For groups exposed to greater delays, Google saw a greater reduction in search activity.  Just a 2% overall slowdown induced a 2% drop in searches per user.

And if users’ propensity to search is hurt by a slower system, their propensity to do other things – like, for example, pull out their wallet and buy something – is hurt just as much.  Bing found that a 2 second delay in search results cut revenue per user by 4.3%.  For the amount of traffic that sites like Bing and Google tend to see, that’s a very large number.

Without question, then, this is an important result for the consumer-facing web.  But what about the workplace?  How does the responsiveness of a business system affect its usefulness?  Unfortunately, I’m not aware of any similar studies in an enterprise context.

But I certainly think they should be attempted.  Any enterprise has internal systems that are strategically important to its operation, as a whole, but are also user-driven.  That is, they can’t simply be mandated into their maximum utility; they naturally have more utility, the more that employees embrace them.  Examples of these kinds of systems:

  • Time and activity reporting
  • Internal databases, wikis, or other knowledge repositories
  • Issue reporting and tracking (whether software or human ‘issues’)
  • Surveys or internal knowledge markets
  • Code repositories and build systems
  • Software-based security systems of all kinds

In each case, it’s in the interest of the company that their workers use these systems early, often, enthusiastically, skillfully, and in great numbers.   This is the only way to extract maximum utility from these systems. And if they fall short, it actually hurts the company more than it hurts the employee.

The A/B tests from Google, Bing, and others show that slow or unresponsive systems have a cost.  There’s no reason that business systems, too, shouldn’t try to measure – and minimize –that cost.

Accenture’s Urgent Data Center Server & Storage Consulting Positions:

Email Resumes to: kevin.daprile@accenture.com

To view job descriptions or apply directly:

Consulting Manager for Data Center Migration & Relocation

Apply to: http://bit.ly/9J9n8s

Consultant for Data Center Virtualization Architect

Apply to: http://bit.ly/dtpoFu

Manager of Data Center Optimization – Reston, VA

Apply to: http://bit.ly/9psdnX

Consultant for Data Center Server Technologies – Reston, VA

Apply to: http://bit.ly/9rkKQ9

Sr. Consulting Manager for Data Center Server Technologies

Apply to: http://bit.ly/cT7bYB

Consultant for Data Center Server Technologies – West Coast

Apply to: http://bit.ly/c9Ir5N

Consulting Manger for Data Center Server Technologies West Coast

Apply to: http://bit.ly/a2gchv

Sr. Consulting Manger for Data Center Rationalization

Apply to: http://bit.ly/d45ZuR

Consulting Manager for Data Center Storage Technologies:

Apply to: http://bit.ly/dbszDg

Consultant for Data Center Technology Storage Technologies

Apply to: http://bit.ly/dnY6tR

Consultant for Data Center Database Administrator

Apply to: http://bit.ly/a10dec

Sr. Consulting Manager Data Center Storage Technologies

Apply to: http://bit.ly/cT7bYB

Consultant for Data Center – Unix/Linux Virtualization Architect

Apply to: http://bit.ly/bY8EPI

Consulting Manager Data Center Unix/Linux Virtualization Architect

Apply to: http://bit.ly/bz9hSx

Consulting Manager of Data Center Optimization – West Coast

Apply to: http://bit.ly/by65AQ

Advertisements

Leave a Comment so far
Leave a comment



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s



%d bloggers like this: