IT Operations Management (ITOM)

HP Operations Analytics - Predictive Analytics – Because looks can be deceiving

HP Operations Analytics - Predictive Analytics – Because looks can be deceiving


As an expert you’re expected to find the root cause of a problem in your domain—be it within the application, network, system, or whatever other area you may be in charge of. As you investigate the problem, you gather a lot of information including: different tables, graphs and charts. The assumption is that what you see will help you analyze the problem at hand. The challenge is that what you see, isn’t always what you get. In this blog we’ll share a couple of examples, and show you how the new Predictive Analytics feature in Operations Analytics (OpsA) 2.1 can help.


False Positive

In this first case you can see a rise in the application response time:



Your assumption is “Oops, there’s a problem developing in front of my very eyes”.


You start investigating, and also see a rise in some of your system metrics (e.g. CPU, Memory etc.):


You think to yourself – I really need to look into this and understand what’s causing the problem …

But wait a minute! Is there really a problem?


Let’s use Predictive Analytics and see:


 Using Predictive Analytics we see the performance has shown up and down behavior over time which tells you that this behavior is pretty repetitive. You also see that the response time typically goes up on Sunday. By using this information, we understand that the current performance change is actually part of the normal behavior of the system. If you see metrics that show that this is just a regular seasonal issue, or occurs at certain times etc. Relax – this is just showing normal system behavior etc.


All Looks Good

In this second case you’re tracking your system, and everything looks good.

The application seems to be OK, and response time is pretty constant:



The underlying system seems to be behaving normally too:



You’re lulled into a sense of calm.

But, ask yourself, “Is this the calm before the storm?”


Use Predictive analytics to check your trend:


You now know you have an upward trend in your memory consumption.


Looking at the Predictive Analytics for your application performance you see an ascending trend in your response time too:



These trends are a strong indication that you have a memory leak that’s very slowly affecting performance.

The trends are weak so there is a chance that they could have easily gone unnoticed. Predictive Analytics helped you uncover them, and solve this potential problem before customers were affected. Solving this performance issue before it hit the radar makes you a superhero!


Looks can deceive

We all love visuals, and pictures speak louder than words. However there’s often more to things than meets the eye. Predictive Analytics can helps you uncover these things, and give you the information to make the right decisions quickly.   We have an upcoming webinar that you won't want to miss! Join Gary Brandt, HP Global IT Functional Architect, to learn how HP IT incorporates best operational practices to collect and analyze structured and unstructured data using big data analytics at enterprise scale.


If you have additional questions about operations analysis and the power of Predictive Analytics, the HP Operations Analytics homepage is a great resource.  Here are a few blogs that might interest you as well:

This Blog was written together with Shimrit Yacobi, Operations Big Data Analytics R&D

  • infrastructure management
About the Author