Guavus Blog

All the latest from the world of Guavus

The New Business Imperative for CSPs: Virtualization & Orchestration: (part 2 of 2)

by Naresh Rao, Director, Product Management, Guavus, Inc.

New perspective for a new era: operations analytics

As we examined the previous blog, the move to virtualized networks has brought along its own set of challenges along with benefits. Communication service providers (CSPs) are left with the task to find new tools to meet these challenges and fill the gaps. Operations analytics do just that.

New operations analytics technologies use AI to help operators transition to an automated, real-time, service-oriented approach focused on optimal consumer experience. Unlike traditional technologies that rely on manual rules and explicitly-defined patterns, AI can learn from data and automatically adapt to new information, intuiting connections and relationships to proactively detect anomalies and prescribe the likely solution.

This kind of AI applies real-time data collection, behavioral analytics and reasoning to automate orchestration and assure carrier grade services in a virtualized environment. It quickly analyzes the large number of variables that pertain to the health of the network across all available layers to understand the user experience.

AI finds new connections in the following ways:

  • Contextualizes: Bridges the silos of data and interprets the data in the context of all relevant information, including performance metrics, alarms, topology, customer experience indicators, etc. across physical and virtual domains. Re-establishes the server-application relationship that has been broken down due to the application being spread across many virtual machines.
  • Understands: Analyzes data using advanced behavioral modeling to detect anomalies, pinpoints root issues (e.g. is it the RAN’s fault? Or one of the virtual apps?), assesses the impact on customers, and makes recommendations.
  • Automates: Analyzes patterns and automatically makes changes based on learned behavior (e.g. orchestrator automatically switches back and forth between small and large configurations depending on demand).
  • Reasons: Predicts failures by discovering signatures that portend performance issues to enable proactive maintenance.

Meeting the challenges of MANO

Guavus AI-powered analytics address all the new requirements and associated challenges with ETSI NFV Management and Orchestration (MANO), enabling end-to-end orchestration of service provisioning and assurance for hybrid networks that comprise both virtualized SDN/NFV-based components and traditional network technologies.

The Guavus solution is designed to meet the demands of the new systems infrastructure:

  • Real-time management of virtual network functions
  • Policy-enforcement, event-driven orchestration at scale, in real-time
  • Service elasticity – Supporting new VNF service models enabled by SDN/NFV
  • Fully automated closed-control loop between fulfillment and assurance processes

Example: Service Elasticity

Let’s look at the following example of a CSP dynamically adding or shrinking new services to support elasticity.  The operator has deployed virtual packet core network elements and an orchestrator on top to provide elasticity.

Now the operations team sees that there is suddenly a high amount of bearer failures. From the network function virtualization infrastructure (NVFI) logs, it’s quick and easy to identify that the CPU utilization of that particular packet data network gateway (PGW) is running at 90%, and there is a steady increase. Based on standard thresholds set in the system, the operator can immediately instantiate another instance of PGW virtual machine (VM) to reduce the load.

Sounds easy!

Not quite. In reality, the failures may not be because of the high CPU consumption, rather the policy and charging rules function (PCRF) may be having trouble, which would cause the PGW to buffer all the packets and spike the CPU levels.  This may or may not be in the same virtualized domain. So the true need is to quickly correlate information across domains and relate the mobility performance counters with the NFVI layer to identify the appropriate action.

By ingesting and correlating multiple sources of disparate information, Guavus can understand the mobility failures and quickly map the NFVI incidents to the PCRF rather than PGW.  Based on this insight, the suggested remedy is to instantiate a new PCRF instance instead of a initiating a new PGW VM to handle the excess load.   This visibility across physical and virtual assets is what is really needed to quickly identify the true issue and fix it.   Guavus operations analytics enables NFV Management and Orchestration (MANO) to be effective and manage systems in an optimal fashion.

Improving Customer Experience – The Ultimate Goal

The investment by CSPs in new virtualization technologies has multiple goals, including reducing costs and delivering new revenue streams. Delivering a high quality of experience will be critical to the success of these investments, yet the increasing network complexity has contributed to the lack of clarity in assessing the quality of customer experience. Service providers must approach the problem strategically and embed advanced analytics as a key pillar of service operation in virtualized environments.

Real-time operations analytics enable CSPs to assure superior customer experience while maximizing the full benefits offered by virtualized functions. Operations analytics consolidate data sources across physical and virtualized systems and perform detailed analysis, in real-time, using AI to optimize resource allocations, detect anomalies and root issues for quick remediation, reduce alarm noise, enable proactive maintenance.  All of these benefits combined lead to the ultimate goal, improving the customer experience.

To see Part 1 of this blog, please click here.


Image attribution: