Metric Manager, AI derived from Watson
In previous articles, Compose IT has mentioned Metric Manager (formerly called Predictive Insights) from IBM and it is time to tell you about our experiences so far.
Initially, we wanted to find out what value Metric Manager could add to our customers and their IT environments and to do that we set up Metric Manager in our own lab and started testing with production data. In parallel, we also had the fantastic opportunity to visit IBM in Cork, Ireland, where we met the people who developed Metric Manager.
Metric Manager comes originally from Watson, IBM’s artificial intelligence (AI) research program. The goal is to use AI to extract value from indicator values. Metric Manager can read time series of indicators from several different types of sources and uses a number of selected algorithms from Watson to analyze the input. The algorithms work together and independently to create dynamic patterns that describe the normal behavior of the time series. The patterns are created in real time and adjust as the behavior of the indicators changes. By constantly continuing to learn the normal behavior of the indicators, it becomes possible to obtain value that cannot be extracted with static thresholds. Because Metric Manager knows what is normal for all times, it can alert as soon as a deviation occurs, thus eliminating the risk of discovering an error too late or receiving a false alarm.
Knowing the normal behavior of each indicator also makes it possible to find relationships between indicators. Metric Manager finds relationships by finding the indicators that have repeatedly influenced each other, e.g. indicator value 1 increases while indicator value 2 decreases. The ability to find relationships makes it possible to reduce the number of alarms as the related alarms are merged into one. It also facilitates troubleshooting as all affected parties are known.
Metric Manager’s knowledge of the normal behavior of indicators combined with its ability to find relationships allows it to live up to its name and actually predict the future. It can predict how an indicator is likely to behave in the next 6-12 intervals. It can also predict whether a deviating indicator can cause a deviation in another indicator. In summary, Metric Manager gives you the ability to act before a problem actually becomes a problem. Metric Manager gives you the ability to be proactive instead of reactive.
Products like Metric Manager offer, properly used, fantastic value. To create this value, one must be aware of the challenges that come with using machine learning and AI.
Data is the foundation of AI because everything it knows is based on the information it receives. Therefore, it is also the data and above all the quality of the data that pose the biggest challenges. It is important to provide AI with good quality data that reflects reality. If you give it incorrect, corrupt or biased information, it affects the results because the AI learns the patterns based on the information over time. The longer it is trained on incorrect information, the greater the effect it will have on the outcome. The quality of the result will not be better than the quality of the data it is based on. Good data quality is the most important component to a good result, but not the only one. The information given to the AI machine can be perfect and still give a bad result if the information is not analyzed properly. The algorithms must be applied in the right way because some algorithms are made for a certain type of data while others have a general application. It can be difficult to know how the algorithms should be applied as some algorithms work together and others work independently. Metric Manager solves this better than the competition and makes the decision for us by automatically knowing which algorithms to apply. This makes configuring Metric Manager simple and smooth while giving you more time to focus on what’s important.
Now you’re probably wondering how the lab we set up went, did we find anything interesting? The short answer is YES. We ran both CSV files with historical data and streamed JSON objects with real-time data in the lab. The CSV files and the JSON objects contained two different types of cases. In the one case, we thought that Metric Manager would be the right solution, but it turned out to be difficult to apply. In the second case, Metric Manager was valuable and found potential problems.
More information about our experiences and how our work with Metric Manager is progressing is coming here so stay tuned!
