Automation of Network Services and Operations — Part 1

CSPs have to develop solutions for automating the Network “Services” driven by Intents in addition to the automation of internal Network “Operations” (NetOps) driven by Policies so that the cloud like flexibility, cost optimization, efficiency and resiliency can be achieved end-to-end for delivering the digital connectivity services for B2B, B2B2X and B2C usecases based on 5G and other Network technologies.

Introduction

CSPs have evolved a lot with automation of Network Operations using IaC (Infrastructure as Code) Python scripts, Ansible Playbooks, etc which has enabled the efficiency, simplicity and reduced the operations overhead of managing and maintaining the Networks supported by the various interfaces exposed by the different vendor NEs and EMs. Alongside, SDNs have introduced the software based control of different network domains driven by Policy based automation (PaC — Policy as Code). All this automation of Network operations has created the benefits for internal network management but that’s not enough for CSPs going forward.

With the digital transformation spanning out across the industries enterprises are asking for simple, dynamic, flexible, scalable, ondemand, customized and personalized network services as required for the dynamic needs of their businesses and market conditions. Enterprises are expecting CSPs to enable the Agility for their businesses and needs. Customers want to spend their resources, time and cognitive load for their core business rather than the IT and Network technology. Enterprises are looking for Agility, Speed, Reliability and Flexibility in Network or Connectivity Services in the same way as they consume the Cloud Computing services and CSPs have to enable that and tap the massive potential opportunity before the new & (non-telco) Technology players capture the market.

In order to optimize the network resources, CSP could sell services to their customers only when they need it. One example of such concept is the connectivity between branch offices — today the business users plan a specific capacity that is designed to absorb the possible traffic peaks but probably they do not always utilize such capacity although they pay for it.
With NFV and associated Network Service Automation, a CSP can provide a Web Portal to their business users where they can order on the fly more capacity for specific needs and only for a specific amount of time and hence the CSP can allow to activate a service (e.g. video conference) only when needed. Or a business user can decide to enable an Hosted PABX or modifying the configuration of an existing one to include new branch offices.

Telco Networks are increasingly becoming sophisticated and complex as one hand the existing legacy networks have to be managed while stepping forward to build new networks and capabilities required for 5G services and products.

5G provides uRLLC (ultra-Reliable Low Latency Communication), eMBB (enhanced Mobile Broadband) and mMMTC (massive Machine to Machine Communication) with which it will enable many digital realities and revolution. Hence new revenue streams will open up for Telcos with various new digital products and services for B2C, B2B and B2B2X customers such as V2V, V2X, AR/VR, eSports, UHD Video Services, OTT Services, Smart Home, Smart Industry, Smart Cities, Smart Logistics, Smart Transport, Smart Stadiums, Robotics, Edge Computing, Drones, IoT, IIoT, etc. with densely connected devices and whole new digital world going forward.

Virtual Networks, Intent Based Networks, vRAN (Virtualized RAN), Core Networks Virtualization (mainly for 5G SA), managing hybrid networks (legacy and new), Software Defined Networks, Virtual Network Functions, Networks Automation, cloud native networks, etc are essential for 5G networks and that will add much complexity to the existing networks build and operations.

Inspite of the complexity, the new network architecture, operating model and technologies will increase efficiency, introduce flexibility, future-ready and provide greater experiences and services to end customers.

Challenges & Barriers for Network Services Automation

Managing Virtual Networks for 5G is complex than existing/legacy Physical Networks where the virtual networks are dynamic and need to be adaptable to the changing conditions and environments in real time or near real time or on-demand basis. For eg. change of network functions on the fly (new version or enhancement or upgrade or replace with a different network function altogether) will be much more normal scenario in virtual and intent based network operations which doesn’t exist at all in Physical Networks currently.

Another key challenge for Telcos is all the transformation of the networks has to happen within the limited CapEx and OpEx budget allocations which are being squeezed regularly due to stagnant or worsening profit margins and reducing ARPU revenues for Telcos.

As per the recent survey by tmforum, most of the CSPs interviewed (out of 93 CSP representations) have mentioned that reliance on vendors is the main barrier for the closed loop network automation.

In order to maximize the 5G investments, one-third of the CSPs surveyed have operations priorities such as going cloud-native; removing systems silos; developing a mature 5G specification; reducing time to deliver network services; and removing process silos.

• CSPs that prioritize the adoption of cloud-native technology see the integration of multivendor operations and a lack of in-house skills as the biggest barriers to closed loop automation of the networks
• Those that prioritize removing process silos see their over reliance on network vendors as the most serious hurdle
• Operators wanting to link network planning, configuration and testing first are more concerned about the lack of a clear blueprint than other groups
• CSPs that most want to remove their reliance on network vendors are also, logically, those that perceive their reliance on network vendors as their biggest barrier to closed-loop automation.

New Capabilities to be developed with Network Services Automation

CSPs need to create different capabilities required to manage the Network Functions effectively and efficiently across the network environments (core, transport, access & edge) virtual networks, hybrid networks, hybrid cloud environments, etc.

Portable Network Workloads

With advent of 5G and Edge Computing, there will be a strong need for the Network Functions and workloads to be moved and ported onto different computing & hardware environments dynamically. This will help to run the network functions across the environments as per the demand in respective location or site and meet the scalability and flexibility as required while optimizing the resources utilization as well for cost effectiveness.

Ondemand Execution

For various upcoming usecases such as smart cities, smart roads, smart transport, smart vehicles, etc. Network Functions (especially) in Access Network and Edge Locations need to be executed dynamically when needed and turned off when there is no demand so that the resources in the constrained environments will be utilized effectively for cost and availability as per the demand.

Dynamic Scheduling & Allocation

Network Functions have to be scheduled as per the forecast and plans (eg. sport events in stadiums, live conferences, entertainment shows & festivals, eSport events, VR game shows, etc) and accordingly the resources have to be allocated in respective environments across local RAN sites, edge locations and access network. This will enable the CSPs to utilize the resources cost effectively only when needed as per the schedules and deallocate the resources post the events.

Zero Touch Provisioning, Deployment & Operations

Rapid deployments across environments, networks as code, On-demand instantiation, auto scaling, self-healing, network functions lifecycle management, CI/CD of network functions, etc.

Automated Assurance

AI led real time analytics, predictive analytics, prescriptive analytics, proactive resolutions, cross-domain correlation of events, self-healing, self-optimization, delivering service as per various SLAs (latency of uRLLC, network slices, optimizations, etc), RAN to Core end-to-end analysis, etc will help in assuring the QoS (the correlations coming from monitoring the network devices/functions), QoE (correlations coming from monitoring the users live traffic) and enhanced Customer Experiences as expected.

The platform should also support multi-level correlation for rapid troubleshooting by mapping physical and logical resources to services, functions and geographies. It should provide service oriented navigation and customer impact analysis. A unified network and service management solution can span radio, backhaul, core and cloud with performance and fault management and inventory functions, with customer experience and application monitoring for true end-to-end service quality assurance.

Automated Network Configuration

Modifications and Updates of the network configuration, policies and design along with the network test and validations have to be completely automated end-to-end so that the services can be rolled out quickly in Agile fashion.

Network configuration activities should be consolidated into a unified and automated workflow with discovery, design, homogeneity and scalable compliance management for radio, transport, core and IT networks. Standardized, repeatable and scalable mobile network configuration is needed when deploying new cloud- and microservice-based networks.

Observability

Real-time insights of the networks, continuous real-time monitoring, single pane of glass visibility of network operations, anomaly detection, etc. will enable CSPs to keep track of the various events in the Network Operations and as well as impact on the Network Services and hence prepare enough and avoid the issues and faults proactively.

Single Source of Truth & Common Platform

There should be single source of truth of the Network (inventory, resources, state, configurations, policies, schemas, etc) that is managed in real time for the visibility of the end-to-end network aspects across the environments so that enables for better automation and efficient zero touch operations.

A common platform has to be established for integrating and automating the planning, deployment and operations systems for bringing flexibility, agility and ease of management overall to accelerate and optimize the network roll-outs

Optimization of Network Performance

Real time packet analysis, session tracing, evaluate new network environment/equipment/devices, optimize vRAN/oRAN deployments, identify the network degradations, monitor multi cloud and hybrid cloud workloads and rapidly and dynamically resolve network issues.

FinOps Governance

As the Network functions will be dynamically scheduled, resources are utilized ondemand and environments are constrained, it is absolute necessary to have the FinOps governance so that the zero touch operations will be cost effective and at the same time continuous improvements can be designed for operations efficiencies with economies of scale.

Benefits of Automating Network Services & Operations

• Optimize Network CapEx and RoI
• Visibility of Business Metrics (revenue, churn, RoI, TCO, etc) and appropriate prediction and planning for the relevant business priorities
• Improve Customer Experience & NPS
• Accelerate Time to Market
• Proactive Churn Avoidance
• Efficient Allocation of Resources and Optimzied Usage
• Reduction of OpEx by avoiding the reliance on manual processes and operations

Network Services Automation — Adoption of Industry Standard APIs

With the ever increasing ecosystem significance, CSPs have to make sure the integrations internally within the CSP organization domains and as well externally are made flexible, interoperable, repeatable, adaptable, pluggable and scalable so that network services and operations can be automated efficiently and effectively.

It is highly advisable for CSPs to adopt the industry standard APIs so that it becomes easy and efficient when working with partners, other service providers and other CSPs when providing products and services for the customers.

Network Operations Automation — CI/CD for Network Functions

In order to achieve closed loop automation, DevOps CI/CD is necessary to automate the network functions delivery, testing and operations.

Network Functions & Infrastructure Orchestration

Following picture depicts the ETSI NFV architecture standard that is currently adopted for Network Functions virtualization (VNFs/CNFs) which is the first step taken by CSPs towards the Network Services automation.

For eg. a UTM (universal threat management) firewall will be a large physical or virtual appliance that encompass multiple functions including DDoS protection, IP filtering, VPN provisioning and termination, stateful packet inspection, and anti-virus / spam filtering. They are complicated to troubleshoot and a clear single point of failure for multiple critical functions. UTMs are also by design imperative. This makes them brittle and poor handlers of certain failure and edge cases. UTMs cannot be easily managed via CI/CD and standard DevOps practices.

As cloud native involves decomposing tightly coupled processes into loosely coupled microservices, rearchitecting applications as CNFs often will entail decomposing functional aspects into services running in discrete containers or as separate microservices. Using Cloud Native approach, all of the functions of the UTM could be decomposed as microservices, each functioning as its own application running in it is own container (or containers) called as CNFs. These services could be controlled and monitored from a single management plane, giving the operator the same visibility into what is happening and the same ability to manage the composed whole, but offering additional granularity on service performance. This should also allow for easier bursting to handle outlier events (usage/traffic spikes) and for running security infrastructure as an agile pipeline. Additionally, if one microservice fails, other services in the meshed CNFs would remain functional, resulting in improved overall resiliency and failure tolerance. For example, a failing DDoS microservice would not impact IP filtering, anti-virus, or other functionality in the UTM application bundle. Similarly, this architecture will support the cloud native vision of network services running on common infrastructure that can scale and enhance performance and reliability in the most optimal manner while enabling an operational environment to address DevOps models in the most adequate manner.

NetOps Automation — CI/CD for Network Functions

The following picture depicts a CI/CD automation for network functions using Jenkins as a CI and CD orchestration and various other tools for different aspects. This contains a sample set of tools and solutions but there can be different tools and solutions to implement CI/CD for network functions.

CI/CD Flow for Network Functions

As a continuation of this topic, I will extend this with a subsequent articles covering the details of standard APIs integrations for intent based Network Services automation and also to provide a practical sample NetOps CI/CD implementation of VNF/CNFs automation in a private DC or public cloud or public cloud edge location.

Network Automation — Categories

There are various levels and categories of automation is being evolved in Networks which varies as per the domain in context as specified below.

• Automation of Network Equipment Management (PNFs) using tools such as Python, Ansible, etc, protocols such as NETCONF & information models such as YANG
• Cloud Native Automation of Network Functions (VNFs, CNFs) delivery (CI/CD) within the domain (Core, Transport, Edge & RAN)using framework such as OSM MANO and as well as Cloud Native Frameworks such as Nephio
• Automation of LCM & End-to-end Orchestration of Network Services across the domains (RAN, Transport & Core) and across Central Office, Cloud, Edge and regional Datacenters using frameworks such as ONAP Kohn
• Automation of Network Services for different requirements of various usecases (eg. Media Streaming, Media Broadcasting, Gaming, Smart Cities, Smart Stadiums, Smart Manufacturing, Smart Health & Wellness, IoT, Drones, etc) with hybrid and/or multi networks (FTTx, 5G NSA, 5G SA, FWA, Wifi, NB-IoT, LTM, P5G, etc) leveraging the standards and reference specifications such as 3GPP, 5G MAG, etc.
• End-to-end Intent Driven Automation for Connectivity as a Service or Network as a Service (eg. Zero Touch Provisioning, Service Fulfillment, Network Exposure, Partner Integrations, Flexible Billing, etc) using APIs such as tmforum Open APIs, MEF APIs, CAMARA APIs, etc.
• Closed Loop Network Operations & Assurance (for QoS & QoE) automation using AI/ML

In the subsequent series of this article, detailed exploration will be provided for each of the network automation categories mentioned above.

References

• Cloud Native Telco Workloads (CNFs): https://github.com/cncf/telecom-user-group/blob/master/whitepaper/cloud_native_thinking_for_telecommunications.md
• Examples of CNFs: https://github.com/cncf/cnf-testsuite/blob/main/EXAMPLE-CNFS.md
• CNF Test Suite: https://github.com/cncf/cnf-testsuite
• CNF Test Bed: https://github.com/cncf/cnf-testbed
• LF Magma — Open Source Mobile Core: https://magmacore.org/, https://magmacore.org/wavelabs/ & https://www.radtonics.com/
• Cloud Native NFV Orchestration on AWS: https://aws.amazon.com/blogs/industries/cloud-native-nfv-orchestration-on-aws/
• AWS Whitepaper on 5G Networks CI/CD: https://docs.aws.amazon.com/whitepapers/latest/cicd_for_5g_networks_on_aws/cicd_for_5g_networks_on_aws.html
• CI-CD for Network Functions on AWS: https://docs.aws.amazon.com/whitepapers/latest/cicd_for_5g_networks_on_aws/infrastructure-deployment.html
• Network Lifecycle Automation for Optimizing 5G Investments: https://inform.tmforum.org/research-and-analysis/reports/network-lifecycle-automation-can-csps-optimize-5g-investment/