Without applied intelligence, data is merely a distraction. Raw data, which is pervasive in modern enterprises, is often funneled into an array of complex tools. Even then, it typically requires an exercise in technical origami to shape and fold the raw data to get what you really want: actionable intelligence
Today, we announced our new Infoblox Reporting and Analytics platform, which our customers can set up in as little as 20 minutes and see immediate value. With more than 80 pre-built reports and dashboards, we do the origami for you. Use our “what you see is what you get” graphical pivot builder, tell the system what you want with simple drop-down menus, and, within seconds, you have the data you want and the intelligence you need. Easily customize existing reports or build your own. We do all the creasing and folding behind the scenes.
Speaking of intelligence, the Advanced Technology Group at Infoblox has devised a new way of applying predictive intelligence to a network’s DNS and DHCP data, to improve the accuracy of forward-looking predictions and minimize the margin of error. This patent-pending algorithm, called the Exponential Moving Maximum (EMM), acts as a pre-filter to provide enhanced predictive analytics. Before I explain EMM in greater detail, let’s first review the most common methods of predictive analytics on time series data:
- Simple moving average is the most basic form of predictive analytics, which functions by taking an average of the last n entries to create an “average” prediction. Unfortunately, this does not take trend or seasonality into account.
- Exponential smoothing, such as the Holt-Winters method, is a more complex moving average which takes multiple factors into account such as trends and seasonality of data. While effective, exponential smoothing is typically biased towards recent data.
- Autoregressive moving average is an advanced predictive method that can reflect autocorrelations and can outperform exponential smoothing when applied to non-volatile data over a long historical period. The big setback with this method is the inherent lack of accuracy when the data is statistically variable.
- Associative rule bases, chart pattern recognition, template matching, neural networks, and k-means clustering algorithms. While highly effective, these approaches are largely inapplicable for real-time applications due to their complexity.
DNS and DHCP data is quite volatile, and thus offers unique challenges for all of these methods. To overcome them, our team devised the EMM filters, which – if you’ll excuse a momentary technical deep drive – compile historical values with a maximum aggregator so that the effect of these values can be taken into account by subsequent values, while applying a magnitude decaying exponential concurrent with time. In simple terms, this allows us to better factor volatile data, such as large spikes in activity, into forward predictions on DNS and DHCP activity that correlate with the actual future state of your essential core network services. This is important, as projecting when you will run out of available DNS queries or DHCP leases is essential for keeping your network highly available.
To learn more on EMM, how to use Infoblox Reporting and Analytics, and new reports and dashboards, stay tuned to the platform’s forum in the Infoblox Experts Community (https://community.infoblox.com/t5/Reporting/bd-p/Reporting).
To see Infoblox Reporting and Analytics in action, take a look at this: