Pauley Perrette from NCIS
Every 21st-century science fiction movie or crime-drama now showcases one of the stars being a geek, one that has access to relevant ‘data’ and can provide the breakthrough in the case. In most cases though, you don’t know what you are looking for. Somewhere in the middle of the show, you understand what to search for. You also don’t know how far back you need to go or which database to search within. Then, an intelligent human/s is needed to interpret the results, correlate with some other input, and derive the result. This person finally alerts another human about an urgent action to be taken. The story is the same in modern-day IT. This is the time for revenge of the nerds, the geeks, the data-analysts.
It was as early as 2006 when a UK mathematician, Clive Humby, proclaimed that ‘data is the new oil’. Of course, we engineers-at-heart know that without data, all you have is an opinion. Here at Extreme, we are focused on unlocking the power of data to provide new experiences to IT administrators and users. This is why we have offered customers unlimited data which means you have access to your data for the lifetime of the service agreement. Unlike others, we don’t store data for a few weeks. We don’t charge you extra to store the data for an extended duration. It is included in your subscription! If you decide to transition away from Extreme, you can export your data.
The vision of utilizing unlimited data is not hard. Wouldn’t it be great to compare the footfall analytics in a retail store trending over the last years? How about the comparison of the network performance between software upgrades? Would you like to compare the traffic lines across bank tellers over time? What about comparing network analytics between device upgrades across multiple months/years? How about comparing the performance since the refresh of your network to the latest product line? Do you want to compare the rogues, the interferers across many quarters? This is why unlimited data matters!
How about tens of terabytes of data, every single day – yes, terabytes. In the words of Andrew Ng, the father of AI, more data provides better training to ML algorithms which in turn improves your product. A better product leads to more users, which in turn leads to more data! In this way, this virtuous circle provides continuous improvement and the best products grow more rapidly.
‘Unlimited data’ is also incredibly useful in predictive analytics. You can establish what is normal behavior and thereby what is anomalous. For example, you can determine how to identify trends that are disturbing, whether a criminal has a history of escalating misdeeds or whether your DNS/Radius server keeps getting slower over time. It is not possible to understand the trends with two weeks of data. You need to observe network trends over long stretches to determine seasonality and transformation over reconfiguration or expansion.
How do we treat the data? After data cleansing, all the Personally Identifiable Information(PII) is removed by our amazing data scientists. Then we start looking for anomalies. Not everything needs AI. That’s right, the correct tool could be simple queries or statistics. Depending on the use case, we then use machine learning algorithms, either unsupervised or supervised. Perhaps the most exciting field of data-driven analytics is the concept of Explainable AI(XAI)! If some vendor were to emphatically state to you, we use artificial intelligence, and recommend you take drastic action such as reconfiguration your network, would you take that at face value? If your data-expert were to blurt out the name of the criminal, would you simply accept it? Of course not, trust has to be earned. This is why we focus on XAI to build trust between the results of the ML/AI algorithms and our customers and the only way to achieve this is through fairness, transparency, and governance. Fairness is achieved through the removal of PII and transparency and governance through an interactive approach with customers.
As you can see from the crime-drama analogy, there is a process for problem-solving. Our product managers spend time crafting the problem statement accurately. They then work hand-in-hand with our data-engineers who utilize the relevant dataset. The engineers thoughtfully use the appropriate tools and algorithms to analyze the data. And finally, our knowledgeable GTAC finally partner with our smart customers to address issues. That’s the recipe for success folks – intelligent humans made smarter by data and machines.