What is normal application performance?

By | | 3 min read

Tag , , , , , ,


Peter Drucker proclaimed: “If you can’t measure it, you can’t manage it.” Do you know what’s “normal” for your mission-critical application? Actually, wait a second–with Halloween having just finished up,  maybe the following Young Frankenstein reference is more appropriate. Whenever I focus on the word “normal,” the first thing that pops into my head (pardon the pun) is that famous scene from Young Frankenstein:


IGOR: I’m almost sure that was the name.

DR. FREDERICK FRANKENSTEIN: [chuckles] Are you saying that I put an abnormal brain into a seven and a half foot long, fifty-four inch wide GORILLA?

[grabs Igor and starts throttling him]

DR. FREDERICK FRANKENSTEIN: Is that what you’re telling me?

But I digress.  Most end-users of a site have a basic understanding of application performance as a result of historical use.  When applications display abnormal performance, something is wrong.  Following the discovery of a problem, a decision to continue can be based on “attachment” to the respective application. The following may take place:

  • Leave this web site and never return?
  • “Suffer” through slow performance or return later
  • Patiently wait because the users are curious and in a hurry
  • Impatiently wait because “You gotta see this”

No matter the reason, the result will likely be negative. Furthermore, these frustrated users may elect to tell their friends, neighbors, and coworkers about their experience. Negative perception of a site’s performance can spread like a Santa Ana-fueled California wildfire.

This takes us back to the notion of understanding what is normal. Again: “You manage it if you can’t measure it.”  There is no way to manage user expectations if there is NO knowledge of abnormal performance.  If one is  aware of bad performance, steps can be taken to improve it.

Smart operations and development teams understand the critical need of  performance monitoring and management.  This means they understand what is normal during specific periods period of time.  This involves baselining performance of the applications most critical business transactions. Baselines come in many stripes:

  • Daily
  • Weekly
  • Monthly
  • Seasonally

For example, a corporate portal would typically have a weekly usage pattern.  Users come in on Monday morning, start up their computers, and start working. Fewer users may might show up on Tuesday. In addition, there may also be a spike around 1:00 p.m. as people return from lunch.  For retail applications, “Black Friday” is a huge day and the days that follow. Holiday sales can account for 40% of annual revenue and up to 80% of profits.  U.S. tax season runs from after Christmas to April 15.

Monitoring and Baselining are a critical component of managing your environment and there is no better place to monitor applications than production.  The uses are real, and the results are real. Being proactive in production and monitoring applications is becoming the norm in regards to remaining competitive–as is understanding that all environments are important but production is key. In order to run in production,  a low overhead solution is required; if overhead is experienced, it will skew results and impact performance.

AppDynamics provides two of these key features for an effective APM solution: low overhead and baselining.  AppDynamics is the culmination of all the lessons learned over the past 12 years since creation of Wily Introscope back the early 2000s.  Try us out–I believe you will be pleasantly surprised, and for the first time in recent years, you and your application may get to feel normal.