Cloud Source Blog
In This HP Cloud Source Blog, HP Expert, Christian Verstraete will examine cloud computing challenges, discuss practical approaches to cloud computing and suggest realistic solutions.

Big Data, Big Data, shouldn't we rather talk about information?

big data 2 (1024x720) (640x450).jpgOne of the most used words in 2013 was definitely “Big Data”. It came up everywhere. It’s the new hype. The Google trends graph clearly demonstrates it. And even if the interest in the term cloud is much larger, the term “big data” is coming up very quickly.

 

But I have an issue with the term. It includes the word “data”. According to the Oxford Dictionaries, data are facts and statistics collected together for reference or analysis. In other words, data are raw elements that are available for interpretation. Information on the other hand consists in facts provided or learned about something or someone, or are what is conveyed or represented by a particular arrangement or sequence of things. So, in my understanding, where data contains the raw facts, information interprets them to come up with a description of what they mean.

 

Ultimately what the user is looking for is information, an understanding of what is behind the data, and is not interested in the data itself. So, why do we keep focusing on “big data”, shouldn’t we focus on information instead.

 

In a previous post, I described the three types of data the CIO was confronted with. It’s in the analysis of the combination of all three that the basis for decision making can be found. In my mind, there are three levels of decision making should be thought of.

 

Operational decision making

In the enterprise eco-system, in its market things happen all the time. Spotting those and acting responsively is critical for ensuring the continuing satisfaction of the user and the success of the enterprise. But with the amount of data available, finding the right signal is like isolating one note of one instrument while the whole orchestra is playing full force. First, the relevant signals have to be found so they can be monitored. Second, we need to define what can be considered as an incident so we can spot it. To make things even more interesting, only 5 to 10 percent of the available data is structured, the rest is semi-structured or unstructured.

 

So, how do you spot you have a warranty issue? How do you figure out your supply chain starts to be disrupted? How do you identify one of your suppliers is nearing bankruptcy? In some cases a single event may trigger an alarm, but in most cases it is the correlation of multiple signals that follow each other in a short timeframe. Multiple tweets pointing to a similar problem in a product may lead to a warranty issue. But how many is multiple? That is where your understanding of your eco-system and market plays an important role.

 

Operational decision making requires to look at information over short periods of time. It’s all about here and now. The time zone to be considered will differ from company to company and from incident type to incident type. In an ever more digital world, the ones who can figure it out and act quickly will be the winners. They may prevent incidents to happen, limit their consequences or delight their customers. All in all, they will be the winners as they create an unmatched user experience.

 

But to be able to do that, you need a deep understanding of how your eco-system and market works. This requires a different kind of analysis I call Tactical analysis.

 

Tactical Analysis

Operational analysis is all about the here and now. In tactical analysis we are adding the time dimension. We look at how things evolve over time so we can understand the drivers, the factors that influence the specific signals. It’s no longer about acting as a result of an incident, it is rather about monitoring.

 

Here again, we first need to identify the signals we have to monitor. Often teams start monitoring a large amount of signals. Over time they will realize some are trend makers, others are derivative ones. It’s the trend makers that are of interest as they show how things evolve.

 

It’s those trend makers that need analysis. How do they evolve? Is there a trend we can spot? Let me give you an example. Let’s assume the contract of one of our suppliers states an SLA of 98%. Whether this SLA is a raw measure or is calculated by combining two or more signals depends on its definition. But if this is an important supplier, we want to measure that SLA as we planned our supply chain around the 98%. If it goes below 98, we may expect delays in delivery to customers etc.

 

In operational analysis we would put an alarm at the 98% level and trigger some action as soon as the level is reached. But could we do better. From a tactical analysis perspective, we can look at the trend and if we spot a downwards trend, we may want to trigger the supplier, trying to understand what is happening before we reached the fatal 98% level. The supplier is still perfectly within his contractual terms, but we may discover he has an issue with one of his supplier and jointly we may want to look at how to address that. The end customer will never spot a delay, and we will have removed a potential problem.

 

Again, it’s not the data as such that helps us, but the analysis made. It has given us an understanding of what could happen in the future.

 

Obviously, not all incidents are the results of a trend. A design issue in the development of a new product may result in warranty issues, and there is no preceding information available. So, spotting the issue can only be done with operational analysis.

 

I remember however that, years ago, we maintained historical information on the MTBF, mean time between failure, of every component used in our products. When releasing a new product, this allowed us to calculate a theoretical MTBF for the product before its launch. Occasionally it failed completely, but most often gave a good indication of what to expect. Again, it’s by analyzing the data available that we get indications of what could happen.

 

Both operational and tactical analysis work on the available data. In other words, they rely on the past to identify what may happen or what action is required. Is there a way where we can become pro-active? Yes there is, that’s what I call the strategic analysis.

 

Strategic Analysis

To be able to perform strategic analysis, we need an understanding of the dynamics of the eco-system or market surrounding us. So, the strategic analysis starts by reviewing and analyzing the available data to understand the key parameters identifying the behavior of the environment.

 

Where things become different is that those parameters are then integrated in mathematical models allowing us to simulate the behavior of the environment under consideration. Again, past events and incidents are used to tweak the model and ensure its operations is close to reality. Once we have that confirmation, we can use the model to understand how to address potential threats.

 

By defining a number of scenarios, some of which may be extreme, we are now capable of understanding what happens in case these events take place. We have the opportunity to inject what we believe to be the best response and again see the implications of our decision.

 

This provides us with a powerful tool to gain an in-depth understanding of our environment, allowing us to quickly react.

By doing so, we manage to respond to threats and keep our operations undisturbed at minimal cost. We also delight our customers as they experience a service second to none.

 

Conclusion

We talk a lot about big data, but big data gives us the haystack. What we require is uncovering the needles that are to be found in the haystack. That requires the appropriate data to be turned into information. The return comes from that information, not from the data. Sure, you have to collect the data, well do you really? Some yes, but a vast amount of data is available online today. What you have to do is to isolate the relevant elements, analyze them appropriately and take the right decisions. So, shouldn’t big data be called big information? But that might be too close to BI, business intelligence. Is that why people keep referring to big data?

Labels: cloud| CloudSource
Leave a Comment

We encourage you to share your comments on this post. Comments are moderated and will be reviewed
and posted as promptly as possible during regular business hours

To ensure your comment is published, be sure to follow the community guidelines.

Be sure to enter a unique name. You can't reuse a name that's already in use.
Be sure to enter a unique email address. You can't reuse an email address that's already in use.
Type the characters you see in the picture above.Type the words you hear.
Search
Showing results for 
Search instead for 
Do you mean 
About the Author
Christian is responsible for building services focused on advising clients in their move to cloud, particularly from a business process and ...
Featured


Follow Us
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation.