Revolutionising Networking: The Power of Network as a Service (NaaS) and AI Integration
Introduction
We discussed in a recent article AI and NetOps[1] (McGregor, J. (2024) the concept of “Network as a Service” (NaaS) is becoming increasingly relevant in today’s digital landscape.
This transition to NaaS is similar to the shift of using the concept of cloud services, where infrastructure as a service, platform as a service, and software as a service are being adapted to the networking industry (Figure 1)[2]. The industry is moving towards using network product vendors as service providers, rather than the conventional approach of buying products and then managing them as a service independently. By transitioning their infrastructure to a vendor acting as a service provider, organisations can focus on providing service assurance. This can allow the organisations to concentrate on how they will consume the resources for the end users and reduce the effort on the management of network services.
Analysts are predicting by 2030 90% of enterprises will be consuming 25% of network services in the new subscription-based model [3]. Organisations are moving from their traditional network model to incorporating NaaS services. Services that fall into NaaS are offers such as secure access service edge (SASE) that can give you such products as firewall as a service (FWaaS), Secure Web Gateway (SWG), Cloud Access Security Broker (CASB) and Zero Trust Network Access (ZTNA). With organisations now shifting their strategies to NaaS the need for simplicity and insights through technologies such as AI are significant to reduce the workload on the network teams. The move towards AI is being used in NaaS due to the ability to learn and make intelligent decisions. Where AI can fit into NaaS is with the capability of analysing large amounts of data and providing insights with the benefits of optimisation through automation through network services. The trending feature used in NaaS is AI-powered chatbots and virtual assistants enhancing how we interact with networks, as these bots automate tasks and facilitate troubleshooting through a guided knowledge base. With the growth of NaaS and AI, security is adapting to combat the challenges of securing infrastructure by enhancing threat detection, incident mitigation, and network optimisation. There has been a concept of as a service being used within the telecom industry for businesses before, where they would offer certain aspects of the network to manage. The industry is now moving towards the enterprise space with a strategy to go into each enterprise level from small to large. This article will explore the benefits of providing a subscription-based model through NaaS — with the assistance of AI — whilst ensuring a secure infrastructure.
Delivering value with Network as a Service (NaaS)
As network management transitions to cloud management platforms, NaaS marks a significant shift in how businesses consume services. This innovative approach to cloud networking management outperforms the conventional method of maintaining physical network infrastructure. Instead, organisations are choosing subscription-based services offered by third-party vendors.
There are many benefits to using NaaS:
Cost-Efficiency:
With NaaS, businesses can free themselves from the financial burden of the upfront investment and management costs of physical network hardware. By adopting a pay-as-you-go model companies align their cost with their usage and can thus, focus their resources on innovation, optimisation and reducing their expenses.
Scalability on Demand:
NaaS is predominantly cloud technology; companies can therefore take advantage of its agility to enable organisations to adjust their network resources in real-time, scaling them up or down as needed. This flexibility enables them to meet growing demand without the logistical challenges that come with physical hardware adjustments.
However, some businesses may prefer to have more control over their infrastructure and may not rely fully on cloud-based NaaS. With on-premises NaaS, organisations can deploy network resources on their hardware, located on-site or in a colocation data centre. These could be virtual or dedicated hardware depending on the resource that is needed but the virtual offering can give you more flexibility compared to the benchmark of a dedicated piece of hardware. There are methods of using a hybrid model of NaaS, giving you the option of certain NaaS services being cloud-based such as Security Service Edge (SSE) solutions. This can be more suitable for certain use cases, with one example being in the Defence sector where they might require more on-premises control over the infrastructure than cloud-based.
Global Accessibility:
NaaS management creates a virtual network across geographically distributed network infrastructure. This translates to enhanced connectivity and greater user experience for both internal teams and external clients that can be spread across various locations. Workers do not have to worry about where they are accessing resources either via home or office. They can use this as a service model to provision new offices more rapidly or move to global coverage quicker than conventional means.
Regular Updates:
NaaS providers continuously update and enhance their services, ensuring that organisations remain at the forefront of networking technologies. This frees them from the cycle of frequent hardware upgrades. This is a major drain on resources whereas the operations team does not have to concentrate on bug fixes or security patches and can concentrate on the CAPEX initiatives.
Core Focus Reinforcement:
By entrusting network infrastructure management to external experts, companies can invest more in their core competencies and strategic initiatives, boosting overall productivity. This allows organisations to focus efforts on differentiating activities. Administrators can now use suites of managed network services available alongside integration to other cloud native services (security, systems management, identity etc.). This also has a knock-on effect on the speed of extending the network, public or private.
However, there are some challenges:
Vendor Dependency
NaaS requires a greater dependency on service providers for their management and support. This can lead to concerns of vendor lock-in and limited control over bespoke changes such as configurations and challenges around provider downtimes.
Multi-Vendor Complexity
There may be concerns about short-term costs with savings on initial investment and maintenance, but long-term costs accumulate over time — which could outweigh business benefits. Although using multi-vendor options can help maintain competitiveness, it can also lead to complexities around service assurance. By leveraging the maturing industry of NaaS providers, businesses can now access a wider range of options to accommodate multiple streams of telemetry, providing valuable insights for viewing.
Unlike traditional networks, which are limited to certain features that companies must pay for, NaaS allows businesses to abstract these complexities into the services they need, providing simplified management and access and enabling diverse approaches to their needs. The choice to move to NaaS will depend on the company’s requirements for security and integration capabilities. This allows companies to progress their enterprise through more agile, cost-effective, and technologically advanced solutions. This accessibility allows quicker access to the platforms they need to respond to any threat or challenge detected and use Zero Trust Network authentication and authorisation technologies to access cloud platforms. With NaaS being more prominent in the market the next step to utilise these as a service model is to use newer capabilities such as AI.
First-line support has always been a costly OPEX exercise but with NaaS, the business can abstract the level of complexity of a network underlay and move to overlay networks. Simplifying the networks using AI allows the business to benefit by reducing the levels of support needed.
Next Generation Networking using AI in NaaS
NaaS is changing the way network infrastructure is consumed, and the need for enhanced monitoring and methods to reduce mean time to detection (MTTD) and mean time to repair (MTTR) is becoming more prevalent to resolve. With the complexity of service provisioning and issue resolution being historically lengthy the need for abstraction for NaaS was needed. In the world of network troubleshooting, AI-powered chatbots are a valuable tool that can provide immediate and around-the-clock support. Currently, the operator can interact with the relevant AI chatbot program of their choice but, in the future, there will be progress made: Where chatbots will be designed to detect anomalies and potential problems before they become key issues and use natural language interactions to guide administrators through troubleshooting processes. By analysing network data and historical information, they can continually enhance their responses and contribute to a growing repository of troubleshooting knowledge.
Thanks to their ability to automate routine tasks and offer insights, AI chatbots can help streamline network management and optimisation. They play a key role in diagnosing and resolving network and cyber issues, reducing downtime, and improving overall network performance. With their intuitive interfaces and predictive capabilities, chatbots can augment human expertise and make network troubleshooting more accessible and efficient for administrators at all levels. For example, they might use Zscaler ZDX offering, and engineers can deep dive into the data to build an in-depth analysis of the incident if needed.
Companies in the networking industry are incorporating AI-powered chatbots into their services to improve troubleshooting. BigPanda is one of these companies that can drastically improve the operations of their clients, in one use case the mean time to resolve (MTTR) P3 — P5 incidents was reduced by 75% through automating Jira tickets and auto-remediating selected issues in under a minute[4].
Through these chatbots, users can gain instant access to organisations’ relevant platforms and receive assistance with resolving network issues. This not only streamlines operations but also increases efficiency in the operations department. By utilising AI, companies can diagnose symptoms and provide solutions for common user queries such as slow network speed or difficulty accessing resources. Deloitte further explores troubleshooting and automation using LLMs in the article “LLMs and ChatGPT in Particular to Data Networking”[5].
Chatbots can also analyse historical data and provide recommendations to narrow down the root cause of the issue, which the network administrator can then investigate further. However, there are potential challenges around governance and control of the chatbot’s actions, such as its ability to change configurations, make network changes, and apply security policies. It is important to have proper governance to prevent potential threats from accessing the system through the chatbot’s communication channels.
The next stage of using AI allows AI-powered automation bots that can identify root causes and recommended fixes and then resolve the issue through automated tasks. There is a risk, however, of allowing automated AI bots to take control of resolving issues which could have an unintended knock-on effect. Further, the knowledge base might have gaps in probable causes. NaaS providers would benefit from these services to allow the businesses to ramp up new services when needed and any non-standard configuration or issues occurring will be automatically resolved.
Using workflows and teaching the AI bot, businesses can train the bot to ensure critical infrastructure is only touched when going through vigorous testing. At this moment in time, there might be more of a recommendation and intervention from a human to allow the change until fully trusting the bot. To draw a parallel, let us consider the medical field. Just as a complex medical procedure necessitates the expertise of a seasoned surgeon who has undergone extensive training to treat an ailment or injury, a complex issue in your networks would require a skilled professional who possesses the knowledge and abilities to address the problem efficiently. Similarly, just as medical professionals have used their knowledge to develop vaccines, a network engineer applies their expertise to guide how AI can diagnose and even prevent issues. Being able to take the methods from medical procedures, you can build ‘vaccines’ or prevent tasks for the AI to do to stop such an issue from occurring or learn how to find the issue through steps taken. This is the same approach for security that will benefit from these AI-powered bots and Gen AI. Security with bots will be less impactful at this moment in time but will be of greater importance for NaaS providers in the future. The approach of utilising bots and AI in security will become prominent from NaaS providers giving the organisations the comfort of providing full end-to-end service.
Enhancing Security in NaaS Environments
As we discussed in the previous section, security is one of the most impacted areas by AI and its use in NaaS offerings. Hackers are changing their methods and making new efforts to gain unauthorised access. NaaS provides the integration between network and network security closer together. With offers including SASE as mentioned earlier, you can provide this on-premise or transition to cloud-based security. An area that will be at the forefront of security is the use, of AI which can be a way to learn from these new methods of unauthorised access and combat them in real time.
There are already efforts in the industry with vendors such as Cisco which has produced the “Cisco AI Assistant for Security” which assists customers in making informed decisions, augmenting tool capabilities, and automating complex tasks. 20% of all Cisco Talos Incident Response engagements this year have been in response to ransomware attacks alongside an increase in sophisticated attacks on networking devices by state-sponsored actors[6]. The Cisco assistant has been trained on daily security events across different platforms creating a vast knowledge base, hence it can use this foundation to understand event triage, impact and scope, root cause analysis, and policy design, to close the gap between cybersecurity intent and outcomes. This is an emerging technology with reluctance in the industry to fully trust AI to take full control of acting in real-time. There will be a process where the AI will be learning patterns, then giving recommendations, and providing workflow automation on those recommendations. When the AI platform has been trained and adapted to certain rules of the business then the question will be to use AI fully to conduct that full process itself without human intervention.
Despite some of the challenges, AI has security benefits for organisations:
Enhanced Threat Landscape:
The integration of AI into SecOps has revolutionised threat detection and prevention. AI analyses network data in real-time; it can identify anomalies, seek patterns, and notice potential breaches that may otherwise stay hidden (i.e., data processing errors, noise, fraud, cyber-attacks, or a sudden or systematic shift from previous behaviour). This initiative-taking approach allows companies to intercept and prevent threats, stopping them from escalating into critical breaches.
Behavioural Profiling:
The key to this convergence is the ability of AI algorithms to comprehend trusted user and device behaviours within a network. With this understanding, AI can quickly detect abnormal behaviour such as unauthorised access attempts or covert data collection.
Automated Incident Mitigation:
In the event of a threat detection, AI’s quick judgment can trigger countermeasures promptly. This increases operational agility, reduces response times, and minimises any negative impact caused by cyber-attacks. For example, AI can be used to automate incident response procedures after identifying a phishing campaign. If a phishing attack compromises user accounts, AI can proactively quarantine them to limit the impact on the organisation.
False Positive Mitigation:
At times, traditional security systems create a multitude of false alarms, causing SecOps teams to feel overwhelmed. Nevertheless, AI technology efficiently filters and connects data, significantly minimising false positives, but there must be an effort to standardise infrastructure and resolve many large numbers of issues so that it does not find false alarms through incorrect configuration. This enables SecOps to concentrate on genuine threats and enhance their overall efficiency.
Network Resonance and Optimisation:
The partnership between AI and SecOps extends beyond just ensuring security and helps to improve network performance. With the analytical capabilities of AI, it can identify areas of inefficiency, streamline network traffic, and anticipate potential performance problems. These efforts lead to a stronger network and enhanced experiences for end-users.
To summarise, the integration of AI and SecOps provides a shield for the networking industry against attacks, while also offering greater flexibility and adaptability to combat emerging threats. With the help of AI and SecOps, companies can enhance their networking capabilities and strengthen security measures through features like Secure Access Service Edge (SASE) and Zero Trust.
As we conclude our article on NAAS, we invite you to join us in our next article. We will investigate the ins and outs of networks with technical details on how networks work with the underlay, overlays, tunnels, and future ideas around what is next in the industry.
Conclusion
In conclusion, NaaS is revolutionising the networking industry by offering a subscription-based model that allows organisations to focus on providing reliable services while reducing the burden of network management. AI plays a crucial role in NaaS, enabling the analysis of large volumes of data to optimise network performance through automation. AI-powered chatbots and virtual assistants are gaining popularity in NaaS, automating tasks, and assisting with troubleshooting through guided knowledge bases. These chatbots can interact with network administrators, detect anomalies, and provide recommendations to resolve network issues. Furthermore, the integration of AI and Security Operations (SecOps) in NaaS enhances threat detection, incident mitigation, and network performance optimisation.
The future of NaaS looks promising, with SASE and AI playing a significant role in providing secure and efficient network services. As the industry expands into the enterprise space, NaaS providers are offering comprehensive end-to-end services, providing organisations with the confidence to rely on AI-powered bots and Gen AI for robust security and network management. With the benefits of NaaS and AI, organisations can easily scale their services, automatically resolve configuration issues, and ensure a secure network infrastructure.
References:
[1] McGregor, J. (2024). AI Ops and NetOps Evolution. [online] Deloitte UK Cloud Engineering Blog. Available at: https://medium.com/deloitte-uk-cloud-blog/ai-ops-and-netops-evolution-89641e37f199.
[2] Jamie McGregor. (2024). Cloud and NaaS [Figure 1].
[3] ABI Research. (2021, June 22). 90% of Enterprises Will Adopt Network-as-a-Service by 2030, but Uncertainty Keeps Market Nascent. ABI Research. https://www.abiresearch.com/press/90-of-enterprises-will-adopt-network-as-a-service-by-2030-but-uncertainty-keeps-market-nascent/
[4] Big Panda (2021) How BigPanda helped bring order to an online gaming multiverse. (n.d.). Available at: https://www.bigpanda.io/wp-content/uploads/2021/09/ECSG_BD_0821-online-gaming-multiverse.pdf
[5] Tsyganova, A. (2024). Large language models and conversational AI in Data Networking. [online] Deloitte UK Cloud Engineering Blog. Available at: https://medium.com/deloitte-uk-cloud-blog/large-language-models-and-conversational-ai-in-data-networking-3ba61cbfc63c.
[6] Cisco Talos Blog. (2023). Quarterly Report: Incident Response Trends in Q1 2023. [online] Available at: https://blog.talosintelligence.com/quarterly-report-incident-response-trends-in-q1-2023/.
Note: This article speaks only to my personal views/experiences, is not published on behalf of Deloitte LLP and associated firms and does not constitute professional or legal advice. All product names, logos, and brands are the property of their respective owners. All company, product and service names used in this website are for identification purposes only. Use of these names, logos, and brands does not imply endorsement.
