The enterprise network is undergoing a transformation driven by both mobility (Wi-Fi and wide-area wireless services,...
especially LTE) and software-defined networking (SDN). Couple the above with cloud-based management capabilities and we pretty much have the network architecture that will carry us for the next decade -- but with one little problem remaining.
That problem is how to manage growth. Demands for network services will continue to increase, generated by both users (more mobile users, armed with more devices) and applications (cloud-based, streaming video and the Internet of Things).
Network analytics boost reliability
Reliability and capacity will be key: No network means no productivity. And all of these requirements can make it very hard indeed to understand evolving patterns of demand and to plan accordingly. And while you're at it, don't forget to minimize total cost of ownership, which spans both Capex and Opex.
Fortunately, there's an opportunity now emerging that promises (at the very least) to manage, if not eliminate, these challenges: network analytics. In fact, I believe the rise of the network analytics market will be so rapid and dramatic that many networking professionals will wonder how they once had to get along without this set of capabilities.
Just what is analytics? It's sometimes described as the set of tools and techniques applied to big data when one doesn't know what one is looking for. Classical analysis capabilities are appropriate when one already has an understanding of the nature of the data under consideration -- for example, using a spreadsheet and simple business graphics. But suppose the volume of information under consideration is very large, multidimensional, multivariate and unstructured? It can be close to impossible to find the relationships between causes and effects without tools designed for just this situation. That's where network analytics comes in.
Analytics isn't new. These techniques have been hard at work for years in such fields as mechanical design, molecular and chemical modeling, and the analysis of economic and business data -- for example, in automated stock market trading. The key output of analytics is insight, usually in the form of sophisticated graphics that make it easy for a mere mortal to grasp the relationships noted above quickly and accurately, and to take action that is precise and effective.
How analytics applies to networking
How does analytics apply to networking? Well, first and foremost, networking is in fact a big data problem today. All network elements generate logs and related performance data; traffic flows can be monitored, application requirements recorded, and many more data elements can be used to make operational decisions.
Are users having trouble connecting to the organizational network? Which users, in what locations? Using what devices running what software? Is this problem related to the wireless LAN? The wired network? A particular time of day? A change in console settings? Upgraded firmware? New equipment? Some pathological combination of these, and more? Applying the power of network analytics is the only strategy to deal with these more-variables-than-equations types of problems. Analytics can thus literally (and rapidly) pay for itself, which is why we believe there will be a rapid uptake of tools like these.
Network analytics for the future
In addition to boosting productivity, increasing efficiency, and optimizing the application of precious capital and operating funds, where might analytics in the network take us in the future? Here are just three examples:
- Console automation: Suppose we create a feedback loop between network analytics, in this case bypassing human user input to directly modify management console settings to correct -- or even avoid -- problems? This is a key direction for predictive analytics, which is really a branch of artificial intelligence.
- SDN: Suppose we use analytical insights to modify SDN controller settings, assuring these settings really are in concert with specified and intended policies?
- Cloud-based multiclient services: Finally, suppose we implement network analytics (as will be the case with many network functions in the future, as network functions virtualization becomes more commonly applied) as a multiorganization cloud service? In this case, the analysis of problems noted at one organization might be used to correct or even avoid these problems at another -- automatically, transparently and efficiently.
There's more -- much more. Analytics needs to be on the front burner in IT shops everywhere. Yes, it's that important and valuable. In this case, the use of the term, "major trend" is truly an understatement.
Analytics for network security
Organizational aspects of storage network backup
Understanding traffic analytics