Hughes Network Systems

Frank Kelly, vice president at Hughes Network Systems, LLC (HUGHES), is the chief technology officer for the North American Division, responsible for identifying innovation and technology to improve service effectiveness and efficiency for consumer and enterprise services. In this capacity, he oversees the strategic direction and implementation of machine learning and artificial intelligence, in addition to applying agile development and service delivery techniques and integrating DevOps technologies into Hughes services. Mr. Kelly earned a Master Degree in Information Technology from Hood College, Maryland, with a focus on network management. He also holds a Bachelor of Science Degree in Computer Science from the University of Maryland.

Researchers at the University of Aberdeen in Scotland recently produced data that shows that the performance of artificial intelligence (AI) systems over the last decade has doubled approximately every six months, significantly outpacing the teachings of Moore’s Law (which estimates that computing power doubles every two years). As just one example, Artificial Intelligence for IT Operations (AIOps) has rapidly expanded from anomaly detection and troubleshooting in the data center to monitoring and predicting failure at the edge. The ultimate application of AIOps combines technical and business information to truly transform business operations. But harnessing the power of AI to streamline network operations – especially in a scalable way – requires commitment, talent and a stepwise approach.

Where Are We Now?

In the evolution of AIOps for enterprise networks, mostenterprise IT currently hovers at about an “advanced beginner” to “intermediate”level. It could be said that “beginner”AIOps solutions focus on operational efficiency at the network core or data center, primarily analyzing and predicting network behavior. Within the past couple of years, more advanced early adopters of AIOps graduated from IT systems monitoring to root cause analysis and problem identification and remediation. Enterprises, and the managed network services providers (MSP) that support them, apply AI and ML across infrastructure and application monitoring to identify and potentially address issues autonomously before they affect network performance or user experience. AIOps is even being applied to monitor and autonomously correct issues at the network edge instead of only bolstering network resiliency at the network core.

At this stage of the AI evolution, AIOps that target WAN edge systems—such as routers, SD-WAN devices, and firewalls—optimize data flow across the edge, predict network traffic patterns across the complete route, and identify and triage possible single-point-of-failure edge devices creating “self-healing” networks.

Where Are We Headed?

The next step in the AIOps journey is insight-driven information services. Expanding AIOps applications more broadly across the digital enterprise environment can uncover data-driven insights that enable the business to deploy changes or corrective actions across an increasingly complex network of devices, applications and tools – even those that may not be directly managed by IT or an MSP. For instance, digital experience monitoring (DEM) could detect disruption to an in-store kiosk system while AIOps pinpoints a potential cause. Even if IT does not have direct access to that kiosk system, AIOps arms them with critical context for troubleshooting and remediation.

Effective AIOPs Solutions Don’t Materialize Overnight. They Require Sufficient Raw Material (In the Form of Usable and Validated Data), a Commitment to Continuous Experimentation and the Right Talent

The final stop on the AIOps continuum is outcomebased services – moving beyond just network and operational performance improvements. Scalability in AIOps distribution represents both a challenge and an opportunity at this stage, activated by distributing the algorithms across the network with an eye toward network expansion and growth. To this end, consider the powerful possibilities of including intelligence in distributed devices for localized data collection and local AIOps decisions as another level of AI impact.

Another example is potential application in shadow IT discovery, an important area of focus in this era of remote work and cloud computing. Beyond simply recognizing a new device on a network, the application of AIOps makes it possible to bring the technology automatically under the IT management umbrella, then identify it accurately enough to write proper API interfaces and bring it into configuration management – all through zero touch provisioning.

What do we need to get there?

Despite the great potential of AIOps and the speed with which AI and ML are advancing, many businesses are struggling to translate enthusiasm for these data sciences into action. A recent report commissioned by Dell Technologies reveals that 60 percent of IT decision makers feel their teams do not have a strong strategy to implement AI and AIOps, and 77 prcent feel their organizations lack a culture of innovation.

Effective AIOps solutions don’t materialize overnight. They require sufficient raw material (in the form of usable and validated data), a commitment to continuous experimentation and the right talent. In fact, the quest for a transformative automated AIOps application demands a significant amount of human perspective. Exploratory analysis of data relies on both ML and data scientists to identify and determine the specific metrics essential to customer service agents and engineers. While ML seems an efficient solution, it tries to solve a particular problem based on a set of specific parameters. Conversely, statisticians and data scientists examine raw data without a specific result in mind, instead reviewing the figures for patterns or anomalies.

Behind every AIOps solution stands a diverse team of domain experts that tweak and test AI/ML models extensively and constantly to ensure AIOps success. For example, in the area of network operations, network engineers understand the nuances of ML systems and the necessary AI algorithms to solve a particular problem with accuracy. Meanwhile, nontechnical experts contribute sector-specific knowledge such as the source and usability of datasets, business strategy and operations. These domain experts also play a valuable role in interpreting AIOps solutions, creating “explainable AIOps” for executives and eliciting stronger buy-in and trust from business leaders unfamiliar with the technology.

AIOps can be truly transformative for a business. Advancing from intermediate adoption stages to outcome based AIOps requires a commitment from company leaders to engage and trust their tech talent and embrace stepwise improvement and experimentation that will ultimately pay off in more efficient, secure and streamlined operations.