Hi Nathan,

We are working on getting more roadmap visibility in the community, for all products and solutions, but right now it is still a bit basic as you will see from the link provided by Michiel.

So to answer your question, specifically for AIOps / Augmented Operation:

• Forecasting: DataMiner will automatically evaluate all performance metrics, and attempt forecasting those.  This is visible if you open up a trend graph in DataMiner, there you will see the forecasted values.  Note that as with each capability in the area of Augmented Operation, we are still continuously evolving those capabilities and making these smarter and better.  Forecasting has been around already for a while, but a recent evolution for example is that it will perform different forecasting depending on the time span you are looking at (e.g. forecasting for the next hour, forecasting for the next day, etc.).
• Change Event: DataMiner will automatically evaluate the behavior of performance metrics, and detect changes in behavior (e.g. a metric was always quite steady and now started fluctuating, or a metric had a certain degree of variation and now suddenly that degree of variation has grown).
• Anomaly Detection: DataMiner will automatically try to see if a Change Event is an anomaly or not (e.g. a metric was steady and now suddenly started a steady incline).  Again, this is constantly evolving, and we are currently working on new features to further increase the smarts on that.
• Proactive Alarming: DataMiner will look at values of interest for your metrics (e.g. classic alarm thresholds that you have defined, the absolute maximum or minimum of that metric) and will issue a proactive alarm whenever it feels that based on the behavior of the metric that there is a very realistic chance that it will breach one of those values of interest (e.g. if a metric is steady, and suddenly starts a steady incline, then this capability will give you an alarm quite some time ahead of the value actually reaching the threshold - because based on the behavior of the metric, DataMiner considers the chances of getting an alarm down the road are very high).
• Pattern Detection: whenever you see a specific pattern in a performance graph, e.g. an anomaly detected by DataMiner or maybe a pattern that caught your interest, you can put a label on the pattern (e.g. to document that this was due to a traffic congestion).  If you do that, DataMiner will based on pattern matching start looking for similar patterns, and will automatically label those for you.  Furthermore, you can also opt to generate an alarm if DataMiner finds a similar pattern.  Again, we still continuously work on this, and we are further expanding this capability.  But this is all about being able to identify things happening in your operation based on patterns and behavior of metrics (rather than simple threshold breaches and absolute values of metrics).
• Incident Identification: this is all about replacing manually pre-defined correlation and root-cause analysis logic with automated grouping of alarms (i.e. identifying incidents) based on learning from the underlying data itself.  We also have a first version out for this, which bundles alarms based on an assessment of when the alarms came in, on which elements and views these are coming in, the location / rack of those elements, the past behavior of those specific individual alarms (e.g. did they occur in the past, when and how did they behave), and other information.  Again, also this capability is currently being further evolved, and one feature recently added was the ability to include custom properties that you have on your elements (i.e. you can ask DataMiner to take that data into account to make a decision for Incident Identification).

Hope this answers the question.  Feel free to get in touch to go through some of this in detail.  And in the meanwhile we're also working to bring more structured roadmap information to the community, for this innovation track as well as for any of our other solutions.

Regards

Ben