Without a doubt, you can expect artificial intelligence (AI) foundation models to play an important role in many organizations’ strategies in the next three to five years. However, if executives are to achieve the business improvements they expect from AI, their teams must ensure optimized network operations and infrastructure.
Revealed in their latest research, EMA concluded that over 40% say that a “shortage of skilled personnel” is the biggest challenge to successful network operations. With the complexity of networks and network delivery having no end in sight, there is no choice but for vendors to help simplify the day-to-day operations for overburdened network teams, and AI is the key.
Effective AI
Data is key to AI working effectively. If you think about it, network operations troubleshooting is not much different than cooking. It involves steps that include following the recipe (guidelines), collecting the ingredients (network data), blending them (correlation and analysis of collected data) and producing a tasty end dish (pinpointing the root cause of the issue). But if you are missing any of the required ingredients, don’t expect the meal to turn out like the recipe suggested it would.
Teams must harness intelligence from alarms, faults, network flows, performance data, network configurations, DNS, logs and more. It is only with this complete operational visibility that AI can help teams make fast, intelligent triage decisions and reconcile root causes faster. By collecting all the relevant data formats and data streams, AI can use this data to help network operations teams establish a pinpoint focus on what matters. AI can empower network teams while reducing escalations — not prolonging triage times with inaccurate predictions and analysis. Missing one ingredient of this recipe could mean that the root cause is hidden from your network operation center (NOC) for hours…or even days.
So, if your network is slow due to congestion or dropping data packets that AI relies on can you trust the answer AI is giving you?
Fortunately, when it comes to AI-powered network operations, we are not starting from square one. Analytic capabilities have existed in network observability solutions for decades now. To meet the intensifying network demands introduced by AI and ensure resilient network delivery, you need the following capabilities:
- Establish a unified data model: Teams must collect the most relevant network metric data available from different vendors and domains, including traditional, software-defined and even externally managed networks of ISPs and cloud providers. Teams need a solution that is smart enough to collect, normalize and correlate this data and then present it in intelligent, unified views of global network health.
- Implement topology-based fault isolation: Teams must be able to model the relationships and dependencies of today’s network components, so they can easily identify the culprit device and its effect on neighboring devices.
- Unlock network fault suppression: Solutions should enable operators to isolate the ‘real’ network fault and suppress the alerts generated by downstream network components that are affected.
- Enable performance metric projections: Teams must manage baselines, thresholds and historical data to build accurate projections of the future This enables groups to anticipate problems before they affect network delivery.
- Deploy traffic flow analysis and anomaly detection: Administrators must be able to analyze patterns of network traffic, so they can gain the insights they need to speed troubleshooting, optimize network performance, identify security threats and enhance traffic segregation and routing.
These foundational capabilities should be in any team’s network observability solution and are the key to fast root cause isolation and analysis. We will never be able to get away completely from IT outages, but we can get better at preventing them and limiting their impact on users.
AI Promises Improvements
AI promises massive improvements in our personal and professional lives and business performance. Increasingly advanced AI technologies can provide powerful solutions to difficult problems — including within the sphere of network operations. However, this virtually unlimited potential isn’t assured — even the most fanatical data scientists will tell you AI is not perfect. Start small and leverage the capabilities above to establish proactive management, so that you can ensure your business gets the most out of its AI investments.
