ServiceNow Machine Learning and Time Series Data

Coming to a ServiceNow instance near you.. Machine Learning and Time Series Data! These two very exciting new features will roll to feature releases soon. Could a time machine be far behind?

Machine Learning

With the recent purchase of DX Continuum, ServiceNow is taking a leading role in making machine learning accessible to the enterprise. The PAAS introduction, embedding ML technology in the platform itself, offers advanced analytic and predictive capabilities.

Customers will be able to take advantage of a simplified interface to a centralized ML Platform Service, and codelessly create predictive data models. The resulting models can support a myriad of use cases. Imagine accessing predictively-modeled data via an API call and using it in custom applications to predict choices or selections for users.

Here is the basic flow:

Service Now’s interface is easy to understand, easy to use, and very flexible. It’s a huge improvement over current code- and algorithm- intensive ML libraries. You don’t need to be a data scientist or expert Python Machine Learning programmer to achieve results. This simplified gateway to Machine Learning horsepower is a no-code solution. Making use of the resulting predictive model minimizes coding in most applications.

In practice, there are just three basic steps:

Step 1: Select Historic Data

Complete a ML Solution Definition record in ServiceNow and submit it. Because this record is a template, all relevant fields in the table will be available to form part of the definition. These Input fields will be considered by the Training Service in determining a predictive model for a given Output field.
A target training frequency and default confidence level can be set as well. In addition, override confidence levels can be set ad-hoc depending on other intended audiences for the data.

Step 2: Training Service

The submitted ML Solution Definition is picked up and processed by the centralized ServiceNow Training Service per the prescribed schedule. Progress in the run is shown on a subform in the interface. SN sends back 100% complete when the Training Service completes the model for the customer instance.

Step 3: Use the data

The resulting predictive model exposes an API that may then be used by the customer for a variety of purposes. ServiceNow imagines that customers and their ServiceNow implementation partners will devise many interesting use cases, and suggests the following general types:

•  Build recommendations (for various things)
•  Intelligent routing of work
•  Agent assistance, such as auto-prioritization, auto-categorization, auto-solutioning, etc.

In the Service Desk area, it is easy to see how this could result in a number of improvements:

•  Improving average response times
•  Better routing
•  Better first-call resolution

Metric Base - Time Series DB

Metric Base is ServiceNow's IOT Big Data platform technology, baked into the CMDB.

This application tracks metrics over time, to produce actionable data. Use cases could be anything from determining seasonal spending for retail, patient wait times for healthcare, or a variety of infrastructure resource capacity and performance metrics for IT. This holds the promise of fixing some issues even before they are reported, if orchestration is in place.

With this functionality, you can write platform apps which store and operate on Time Series data. Native in the CMDB, it leverages Content, Data and Relationships. ServiceNow says this makes CMDB the database of IOT.

Features include:

• High Throughput: database performance is optimized to handle inputting 160K values/second to an instance.
• Efficiently Stored: Fast backups, clones and restarts
• Fast ad-hoc, realtime queries: example - Top 5 servers >90% in last 2 hours

3 Steps to use Metric Base:

Step 1: Create a Metric

This is done be completing a Time Series Metric Record in ServiceNow. Specify the table name, the metric field name, the aggregator and re-sampling duration.

Step 2: Collect data for the metric

Data from either external data sources or actual IOT devices can be fed into CMDB tables via either midserver (inside firewall) or  REST (external).
Data feeds can be as frequent as required to build a meaningful data set.

Step 3: Leverage the data to make a decision

a. Group the data easily and display via a number of graphical reporting options to support decisions.
Here is an example showing response times of instances coming from two service providers over time and during use:



b. Access the data itself to connect other functionality on the platform, provide information and triggers for other controllers, actions, etc.

Summary:

ServiceNow is moving forward quickly to provide data functionality that truly takes the platform beyond IT services and into the enterprise. This is an early glimpse at systems that will see a great deal of action in 2017 and beyond; the possibilities are tremendous. We would love to talk to you about your needs in this space.