The $20B Service Degradation Problem: AI to the Rescue
I want the truth
“You want answers?” ... “I want the truth.” ... “You can’t handle the truth!” …It’s the most common statement from every tool or existing solution attempting to obtain the root cause from a complex and convoluted technology stack, especially when introducing virtualization and containerization, in addition to the complexity of underlay and overlay networks.
Just this year alone, major operators in the US, Canada and the UK have faced significant outages of voice and data services due “technical issues” lasting several hours, and even days in some cases.
Based on estimates from Heavy Reading, service degradations and outages are costing network operators almost $20 Billion a year, and this cost is increasing year after year.
The message to the subscriber is always the same: “Our technicians are actively working on stabilizing the network and fully restore the services”, resulting in millions of dollars in fines, churn and operational costs to service providers.
This statement is not far from the truth. A Heavy Reading survey of global service providers found that the current methods of finding the root causes of service issues have not improved substantially over time (below). Another Heavy Reading report found that root cause analysis is the leading challenge faced by CSPs in troubleshooting applications.This results in increased incidence of outages, network degradations, security, performance management, and longer time to repair.
The speed with which networks are evolving and network complexity is increasing, create insurmountable challenges for operators to unravel root causes of any given issue. This creates bottlenecks in CSPs’ ability to transform their operations, improve customer centricity and monetize network investments. Reducing network failures by automating testing procedures and creating a robust knowledgebase of issues and resolutions in pre-production environments will be essential to transformation of CSP business models.
Handling the Truth
It is time to stop managing the symptoms and “bandaid-ing” the network and focus on addressing the root cause. To do so B-Yond has developed the Service Delivery Accelerator (SDA) that uses a powerful and unique AI-powered framework that provides flexibility and velocity on data ingestion and processing.
Our extensive experience working closely with the largest operators around the world has given us unique insights into the network evolution complexity and the introduction of new technology impacts, which we have translated to leveraging AI and created an AI-driven test cycle automation solution to optimize and improve resource utilization, test velocity, scaling, and repeatability.
SDA is a suite of services that enables the operator to automate the test cycles and service roll-out processes which helps to accurately provide Root Cause Analysis (RCA) expediting the time-To-Market (TTM) of new services and platform features. SDA also decreases the Mean-Time-To-Repair/Resolution (MTTR) and also builds a powerful and unique ML knowledge model base that will provide the operator with the perfect tool to evolve as quickly as the telco world is evolving.
❯ Lengthy and manual testing processes
❯ Multiple vendor tools compatibility (methodology, complexity, integration problems)
❯ Delayed Time to Market of new technologies, products, and services
❯ Lack of Cross-platform and Domain Expertise including virtual and containerized infrastructures
❯ Operational troubleshooting involves numerous teams and resources creating collaboration and coordination problems
❯ Reactive issue detection, in some cases after several days
❯ Time consuming troubleshooting using traditional tools
❯ Constant network changes, revisions, and updates are expensive to maintain, operators will experience outages due to network upgrades
❯ Non-existing or unproper RCA documentation and dependency on a small group of subject matter experts
SDA’s ability to meet operator challenges:
❯ Unique metadata driven methodology enabling complex data ingestion correlations and avoiding the need to build costly and non-reusable custom parsers
❯ Cloud agnostic and cloud native implementation
❯ Improved classification over manual testing methodologies with AI-based Root Cause analysis of:
End-to-end call flows spanning multiple platforms and protocols (3GPP, Non-3GPP, NFVI, etc)
Platform-specific testing eg: PGW, SGW, MME, MCPTT AS etc.
Ability to automatically decipher underlying call flow patterns, root causes and reinforcement of the detected pattern using supervised feedback
Project goals applicable to any program and all areas of testing, including Regression Testing, Functional Testing, Service/Platform Testing
Build collective knowledge and expertise overtime that enables closed-loop remediation
High accuracy in failure classification and RCA in short reinforcement cycles
❯ RCA Velocity: Increase RCA Velocity exponentially by creating an extensive AI trained model base
❯ Adaptability: Using known base models to specific services
❯ Operational Cost Optimization: Reducing human effort (from 30% to 80%) in RCA and troubleshooting processes
❯ Accelerate Time-To-Market: For new features and reduce time to new revenue
❯ Reduce Liability on Manual Processes: Enabling true AI-based automation and closed-loop remediation
❯ Knowledge Transfer: Retain collective knowledge on different technologies on RCA and troubleshooting
Twelfth Annual State of the Network Study - VIAVI Solutions
Mobile Network Outages & Service Degradations: A Heavy Reading Survey Analysis