Migration to cloud has led a way to heavily automate the deployment process. Teams rely on deployment automation for not just deploying regular updates to their application, but the underlying cloud infrastructure as well. There are various deployment tools available in the market to set up pipelines for almost everything that we could think of. Faster delivery, less manual efforts, and easier rollbacks are now driving the agenda for Zero Touch Deployments.

What does Zero Touch in Cloud mean?

We would love a cloud environment where workload AWS accounts especially a production account require no console login to design, implement and operate the infrastructure and application resources. The team could have read access to view the resources but that’s as far as they can go. This helps in avoiding human errors such as forgetting to check the resource ARN before modifying/ deleting the resource on AWS CLI command. This happens with a lot of developers. Resolving these issues is what is the idea behind Zero Touch. Using pipelines and IaC (Infra As Code) tools, it becomes easier to apply it practically.

zero-touch-cloud-deployment

In picture (a), the IAM role “Shared-Deployment-Role” in the “Shared Deployment” account is assuming IAM roles in the workload accounts to deploy resources. The workload accounts could have additional roles to allow users to assume and login into a specific account. Users may have read-only access in Prod account to view services and resources. The “Deployment-Role” in each workload account is created along with the initial infrastructure layer using the IaC tool (AWS CloudFormation/ Terraform/ AWS CDK) and Pipelines (CodePipeline/ GitLab/ Jenkins/ BitBucket). AWS CodePipeline is configured in the Shared Deployment account and IaC templates are stored in the AWS CodeCommit repository for version control.

zero-touch-cloud-deployment

Picture (b) gives a high-level understanding of hoe Application deployment and Infrastructure deployment pipelines would look in AWS Cloud.

Infrastructure Layer:

Using CloudFormation templates, CodeBuild and CodePipeline; we deploy resources like and are not limited to IAM roles for deployment, VPC, Subnets, Transit Gateway/ Attachments, and Route53 hosted zone(s). These services and resources are necessary to deploy and launch the application. The resource ID/ ARN values are stored in Parameter Store for consumption by IaC templates for the application. Parameter Store helps in developing re-usable IaC templates. How? The answer is to create Parameter Store keys with the same name across all the workload accounts and allow Infrastructure templates to update the values dynamically. Deployment of the infrastructure layer is generally managed by the organization’s IT team with approved AWS services and the organization’s cloud best practices.

Application Layer:

Every application in an organization can differ in the services required to host it in the cloud. Application developers or DevOps teams can choose any one or combination of approved CI/CD and IaC tools to design and host the application in workload accounts. Teams can leverage CodePipeline, CodeBuild, CodeDeploy in Shared Deployment account to build and deploy applications in workload accounts by assuming respective “Deployment” roles. Remember that the IT team had created parameters that hold resource id(s)/ ARN(s) of resources that could be consumed by application templates. The Agile model for development, test, and deploying application templates are encouraged to be adopted ensuring only clean and tested code/template(s) go into Production.

Conclusion:

There is no one “the best” way of designing infra and application deployment. Size, complexity, cost, and time could determine what is optimal. A Zero Touch Cloud Deployment strategy can comprise various permutations and combinations of infra and application components. However, the motive behind the approach could help in minimizing human errors and many sleepless nights.

DevOps is a term that is not new for the software world. However, it is certainly the magical wand which has really sped up the digital transformation. In a sense, the entire SaaS products story is written with the help of DevOps . In today’s VUCA world, digital services aren’t simply nice to have but are a basic expectation from consumers and enterprise customers alike. In the whole digital transformation journey DevOps clearly aligns well with the business goals, ensuring that the experiences they deliver form a seamless and customer-delighting part of the entire journey.

Continuous delivery and integration with magnificent tools have allowed the companies to create entire products as individual chunks. These individual chunks of functionality, captured by user stories, can be developed, and deployed into production in a day or two, not in weeks or months. That has really changed the game while we look at product development.

The Product Led Approach (PLA) driven by DevOps has created a culture in which the final goal has converted into the delivery of a fixed set of requirements, on-time, and on-budget scenarios. Scripts that can set up the entire deployment infrastructure, including software-defined networking, are managed just like the source code of the services running on them. Business-centric services that can evolve quickly and independently, combined with frequent and reliable releases, finally put the old dream of reusable and re-combinable components in reach for the companies.

How DevOps can help in Digital Transformation?

  • Maturity Model: DevOps is the aggregation of cultural philosophies, practices, and gear that will increase an organization’s potential to supply programs and offerings at high velocity. This results in evolving and enhancing merchandise at a quicker tempo than businesses using the conventional software processes. Enterprises are moving from large, monolithic applications to smaller, loosely coupled microservices. This enables clients to act faster, better adapt to changing markets, and grow more effectively to achieve their business goals. Companies use DevOps continuous delivery practices that help teams take ownership of these services and then release updates faster.
  • Break Organization Silos to Collaborate: DevOps helps in driving the collaborative thought-process and change in mindset. DevOps helps organizations achieve digital transformation by changing the social mindset of the market, cutting off silos, and covering the way for continuous innovation and agile experimentation. With a DevOps model, development and operations teams are no longer “isolated”. In fact, DevOps encourages better communication between the two teams and creates development channels that enable continuous integration. The software problems are identified, resolved and deployed faster.
  • Organize Process around Customers: The increased speed allows companies to better serve their customers and be fair in the marketplace. Processes can be seamlessly designed and finalized based on customers’ business needs, helping them achieve higher value growth. When combined with rich digital telemetry from modern monitoring and observability tools, we end up with a strong knowledge of our systems that helps reduce mean time to recovery (MTTR), allowing teams to really take ownership of production services.
  • Build an experimental mindset: Experimentation is the fundamental need for success in today’s rapidly changing technology stack. DevOps can help create the speed of experimentation at which the business can reliably implement these ideas and launch them into the market to start learning again.
  • DevOps and Cloud: Cloud is part of almost every digital transformation journey. DevOps and cloud are completely synergetic to each other. This powerful combination has empowered the developers to respond to the business needs in near real-time. The latency of software development has become a part of past. The partnership of DevOps with cloud has given rise to a new term generally called ‘CloudOps’. The overall advancement in CloudOps has lowered the total cost of ownership for the organizations. This has made a direct impact not only on the top-line revenue and market share but also on its innovation capabilities and response time. Cloud was created majorly to tackle the challenges of Availability, Scalability and Elasticity goals based on dynamic demand. CloudOps uses the DevOps principles of CI/CD to realize the best practices of high availability by refining and optimizing business processes.

The current AI adoption levels coupled with improved machine learning techniques, is enabling companies to discover and operationalize business insights, previously hidden. Increased data availability and higher processing power is facilitating greater adoption, allowing enhanced methods with more data at lower cost.

Telcos world over are well positioned to make the most of this evolving situation by optimising their business and serving as AI platform enablers for other industries that are expected to spend upwards of $15bn annually on AI-led operations by 2021-22.

Three key trends that will define Telcos’ next decade are the reinvention of asset-light business models, resurgence of Telcos’ enterprise business, and liberation of network infrastructure that will accelerate and create a range of varied business models. Companies that embrace oncoming changes and those that make bold and quick investments will lead the pack and buck the trend.

Ovum’s findings reveal that the Telco market is perceived as second most at risk of disruption only after Healthcare. AI will play a large part in this change as Telcos look to implement technologies that reduce costs through greater automation, improved customer service, and optimized network and traffic management.

Significant improvements through AI will impact Customer Service, and Advertising/marketing operations, translating to improved cash flow for Telcos by 4x over the next 10 years.

Drivers of AI in Telco Ops

Artificial intelligence serves as an extension to a robust operational analytics system. However, whilst technology has been around for long, four powerful forces have accelerated greater adoption of AI.

  • Increased data availability: Available data on average is doubling every twelve months, attributed to increase in connected devices and much higher rate of data from connected devices. Evolved statistical models for decisioning are enabling new ways of managing and analysing very large data sets.
  • Data transformed towards a 360-degree view: Typically, Telco data is not integrated at a deep level, but instead scattered across functional and operational silos. The different data sets can vary widely in terms of quality and depth, rendering some data sets less actionable than the other. AI-powered solutions are helping integrate first & third-party data to maximize the ability of Telcos to achieve genuine 360-degree data view.
  • Increased processing power: Exponential growth of processing capacities is enabling AI solutions to be implemented on higher data volumes at lower cost. Additionally, distributed networks such as cluster computing on cloud have become exponentially more powerful.
  • Improved machine learning capabilities: Commercial interest in AI/ML has been growing strong with technology devoting significant capital to research. As a measure, Google doubled its AI research in the four years 2012-16.

Impact of AI on Telcos

There are opportunities aplenty for Telcos to leverage AI, as summarized below.  (Not exhaustive).

Customer Service & Contact Center Support

The advent of 5G in the telecommunications sector is driving multiple opportunities for organizations. 5G along with accelerated cloud adoption promises to deliver new digital services at a much lower latency through SaaS platforms, OTT services, and cloud based unified communication services, among others.

AI is a key enabler in improving quality of customer experience and quality of service. Telcos’ strategies to monetize data depend in large part on algorithmic intelligence and automation to handle exponential rise in traffic and onboarding new devices and users. Add to that, processing personalized customer service responses.

Key Opportunities

  • Assisted context-aware customer service: For customer service, the major reason why customers call is billing dispute. Using AI, Telcos can analyse user activity across engagement channels, and predict when a user calls in. Identifying intent and tailor their experience to get prompt resolution. Ex., when customer uses app or web to search billing pages, customer search data can flow into Telco Data Lake and finally when customer calls in, customer data from the Data Lake can be used to create context, which is plugged-in during agent call or chat.
  • Predictive maintenance of customer premise equipment: Analysis and insights of contact center notes to better the IVR system. With prior permission, analyse customer behaviour in terms of equipment used to signal a potential problem beforehand and pre-empt corrective action.
  • Conversational AI: Scaling and automating one-to-one conversations to drive down the cost and improve the efficiency of operations and customer service. Contact center first line virtual agent dealing with 90% of routine questions, with emotion sensing capability based on voice and video. Major use cases for implementation in the space of conversational AI are the regular ones that overwhelm contact centers. Ex.;. installation, set-up, troubleshooting and regular maintenance.

Network Optimization

AI is rooted in Telco virtualization of networks, namely, SDN and NFV. A fully NFV enabled network will be controlled by a single NFV orchestrator that dynamically determines critical network operations such as assignment of resources to a network function, provisioning new network nodes, or withdrawing network elements that go underutilized. Traffic will be controlled by a centralized SDN controller augmented by AI that allows for efficient and proactive routing of traffic enabling capacity to be managed effectively, network outages minimized, and faults bypassed. AI can optimize configuration of a Telco network according to dynamic network capacity demands, characteristics of traffic volumes, user behaviour, and other parameters. Network deployments may also be further improved by AI to predict traffic patterns and forecast user trends.

Key Opportunities

  • Telecom network capacity optimization and analytics: Transition to Network Function Virtualization (NFV), Software Defined Network (SDN), and Self-optimizing Network (SON). Self-optimizing network based on traffic information, detect anomalies beforehand and proactively optimize network performance.
  • AI for predictive network congestion and maintenance: Utilizing data, sophisticated algorithms and machine learning techniques based on past data. This helps in close monitoring of equipment, proactively anticipate failures and take corrective action before the customer raises the ticket.
  • Pre-emptive Vulnerability and fraud detection, cyber security helps defend critical network infrastructure from malicious attack.

Marketing Engagement

Understanding  user behaviour enables Telcos to create personalized customer engagements for its customers, creating offers and messages that are contextual and performed in real-time across a wide range of criteria, including personalized pricing plans, service bundles, and marketing messages. Personalized, real-time sales and marketing offers play a central role in Telcos’ data monetization strategies as well as enhance value of customers’ engagements and improve Customer Satisfaction (CSAT) and Net Promoter Score (NPS).

Key Opportunities

  • AI for faster response to personalized sales and marketing triggers, such as creating and tearing down offers. Product bundling recommendations.
  • AI-led customer engagements to replace manual intervention in select sales & marketing business processes.
  • New entries in product catalogues optimized by AI, such as price and size. Deep learning of competition and advertising data can be used to configure these entries.

AI Challenges for Telcos

Telcos face numerous challenges as they consider their next steps with AI, and they would need to keep a watch on the following factors, and they play out.

  • Threat from early movers: Telcos are not the only players looking to leverage AI to improve operations and services to gain competitive edge. Consumer tech. OTT and FANGs of the world are investing heavily in AI, and Telcos fear being left behind.
  • Data privacy: AI can leverage very granular consumer data insights and these capabilities will deepen going forward, to the point where they attract regulatory scrutiny. Telcos should ensure their AI solutions can safeguard data privacy.
  • Realign workforce for new employment opportunities and talent retention:  While the job market will evolve because of efficiencies and productivity introduced by AI-led automation, this will open up new possibilities for the workforce. Our workforce will steer AI initiatives, set goals, provide data and training, and monitor machine activities and performance. Fragmented AI skills in the organization, unclear AI organizational model results in difficulty while attracting and retaining AI talent.
  • Lack of end-to-end operational visibility: Numerous isolated POCs being undertaken by Telcos in the AI space make it difficult to present the comprehensive value and hence lead to insufficient leadership support for projects.
  • Lack of effective change management: Understanding of AI at different levels, unclear impact and value extraction and sporadic AI related training and adoption efforts makes it difficult in percolating changes to all levels uniformly.

How does all this shape the future of Telcos

In a decade from now, a new breed of asset-light carriers with more sustainable businesses will emerge as technology advances and evolving consumer demands reduce costs across operations.

There is more promise AI and digital technologies holds for Telcos, now more than ever before, which now enables asset-light carriers to operate customer-facing functions at drastically lower costs by substituting computer processing power for people. Both consumer and business customers are increasingly comfortable with digitally delivered self-service options, providing an advantage to carriers that build their businesses this way.

What is within the realm of possibility is that carriers could successfully operate with no stores, no call centers and significantly fewer field technicians.

We are seeing early experimentation with this among Telcos globally, where there are digital only wireless services from large incumbents, namely Verizon’s Visible plan in the US, and fixed wireless broadband Starry’s resource-light model in the US. Over the next decade, this shift toward digitalization and automation could set off a chain of similar events in the Telco market.

Proliferation of asset light carriers will deploy this strategy first, and their lower breakeven point will allow them to focus on targeted, smaller slices of the market. These carriers will be either standalone businesses or new arms of existing companies in other sectors that have strong branding and distribution that they can use to push a telecom offering.

Enhanced profitability of these new market entrants will put pressure on traditional Telcos’ price points and revenue, forcing them to respond with their own digital front ends. As per Ovum research, average Net Promoter Score for early adopters has risen by over 20% points, driving up revenue by up to 10%, while lowering customer-facing costs by over 30%.

Emerging technologies deployed in contact center operations cut down incoming customer calls and improve operations cost. As per BCG analysis, deflection rate of 20%-40% for customer calls can result in lowering the contact center cost by 10%-20%.

Taking a deeper look into this potential, AI implementation results in dual benefit for Telcos. On one hand it generates a potential of ~10% increase in revenue, while on the other hand also helps in cutting down the cost by ~15% across the value chain. Few areas gaining traction on the revenue side are lead generation and personalization, managing customer churn rate, upselling and cross selling offerings to customers. Similarly, on the cost side, network optimization, and workforce optimization.

Incedo’s solutions tailored to meet demands of the AI-led Telco world

Incedo has built a set of AI/ML pipelines that can be configured to solve a variety of operation optimization use cases. These pipelines can automate processes, tap into unstructured data sources for intelligence. Our solutions are driven by the following:

  1. Cross-industry techniques: Inspired by use cases across industries – eCommerce Recommendation Engines, Image Search by Google.
  2. Automated model training: Designed with AutoML features for automated learning across ensemble of techniques.
  3. Self-learning in production: Algorithm to identify model performance and run calibration in production. Like Google Maps suggest better routes when found.
  4. Modular design: Developed as Lego blocks for easy integration into existing infrastructures.

Incedo’s AI/ML Pipeline Tailored for Telco applications

AI/ML PipelineObjectiveTechniquesTelco Applications
NLP PipelineTo make sense of free form text dataTopic Modeling – LSA, LDA, RNN
Classification – LDA2Vec, SGD, Random Forest
Sentiment – Stanford Symantec Libraries
  • Analysing Jeopardy Tickets
  • Automated test case generation
Optimization PipelineTo prioritize between actions based on predicted impactMarkov Decision Processes – Markov Chains, Hidden Markov Models, Multi-Armed-Bandit, Reinforcement Learning
  • Emerging Hot-Spots
  • End-Point Placements
  • Personalizing Customer Interactions
Anomaly Detection PipelineTo identify or tag anomalies from a normal behaviorAR Family of Models, Auto Encoders, LSTM, Isolation Trees, Neural Nets
  • Predictive Maintenance
Personalization PipelineTo personalize customer touch points – email, store, website, app, customer servicePropensity Models to estimate intent
Reinforcement Learning to optimize content delivery in the real-time.
  • Web Personalization
  • Cross-Sell/Up-Sell
  • Customer Retention

Incedo is using its AI/ML pipelines to help a US based Tier 1 Telco to enhance its spectrum of network operations and customer service.

  1. Predicting optical network faults 48hrs. in advance to reduce trouble calls or expensive technician dispatches
  2. Automatic redirection of 5G build out issues to the right team for quick resolution.
  3. Automated ticket logging by extracting relevant information from emails, with the help of NLP based email tagging solution.
  4. Prioritize automation opportunities by identifying process gaps and bottlenecks using data mining algorithms.
  5. For customer service, a major reason why customers call is billing dispute. Using AI, Telcos can analyse user activity across engagement channels, and predict when a user calls in. Identifying intent and tailor customer experience to provide prompt resolution.

Cloud Cost Optimization:

Cloud has a decentralized model of consumption where each department or BU has visibility into their cloud consumption thanks to fine grained Account creation and control and billing segregation. The decentralized model has raised costs for organizations exponentially, and often without any control over the spiralling bottom line. Businesses will have to start to get a handle on these costs as cloud usage grows, streamlining the expenditure that they are not utilizing to full effect, and cutting out duplicate spending or unnecessary overheads.

This also provides an opportunity to vendors who can build their tools and services around cloud optimization services.

Hybrid Cloud environments take a big jump with Focus on Automation:

With cloud-native computing [and] container-based workloads gathering steam, enterprises will want to build solutions that take advantage of their on-premise resources and cloud resources equally adding that in some organizations cloud utilization will be driven by specialized circumstances or use cases rather than as the default setting. Serverless architecture driven by containerization and orchestration engines will hybrid cloud approach easier.

Managing complexities arising from multi cloud environment is possible by automation tools along with comprehensive dashboards that provide a holistic view into cloud operations

Delayed migrations due to Insufficient IaaS skills:

According to Gartner “Through 2022, insufficient cloud IaaS skills will delay half of enterprise IT organizations’ migration to the cloud by two years or more. Today’s cloud migration strategies tend more toward “lift-and-shift” than toward modernization or refactoring. However, lift-and-shift projects do not develop native-cloud skills. This is creating a market where service providers cannot train and certify people quickly enough to satisfy the need for skilled cloud professionals”.

To overcome the challenges of this workforce shortage, enterprises looking to migrate workloads to the cloud should work with managed service providers and SIs that have a proven track record of successful migrations within the target industry. These partners must also be willing to quantify and commit to expected costs and potential savings.

Security, reliability, and flexibility drive cloud strategies :

The early convention of cost savings by moving to cloud is no longer the only one although an important one. However security , reliability and flexibility have become key driving factors in mutli-geographic and multi-vendor environment.

Recruitment Fraud Alert