The world is moving towards intelligent networks. Regardless of whether it is the core, the wireless, the data center, the LAN or the WAN, the need for networks to be well aware of the state of traffic they carry, and the nature of content that is being transported is becoming increasingly critical.
Ultra-reliable, ultra-low latency (URLLC) communications and evolved mobile broadband (eMBB), the hallmark services of the upcoming 5G, point to the transition of today’s networks towards becoming super highways, capable of delivering gigabytes of data at millisecond latencies. These super highways will be expected to deliver applications such as remote surgeries, connected cars and smart cities at unprecedented speeds and quality of service. Most importantly, these highways are expected to seamlessly manage disparate types of traffic that demand highly differentiated policy responses.
Key to intelligent networks is network intelligence. As cliché as it sounds, this understanding will massively redefine how networks are operated in an era where gigabit speeds and ultra-low latencies become the norm. Network intelligence relies on network analytics, and network analytics rely on Deep Packet Inspection (DPI)’s traffic identification capabilities. What is DPI? DPI is essentially a packet inspection technology that extracts fine-grained application and metadata information in real-time for traffic management decisions, network analytics and network security. DPI uses heuristic, statistical and behavioural analysis of IP data packets to detect and classify network protocols and applications via the use of continuously updated libraries.
Driving network efficiencies
The information provided by DPI creates firstly, network visibility and secondly, application awareness. In combination, these two capabilities enable network resources such as bandwidth, and network services such as data compression, acceleration and caching to be allocated according to traffic management and prioritization policies. This drives network efficiencies while delivering the expected quality of experience on each application. Not just this, real-time classification of data packets goes a long way in enabling networks to become automated, with advanced capabilities such as self-optimization and self-healing.
DPI traffic identification supports these policies at both an application and network level. At the application level, different applications, based on their speed, latency and mobility requirements receive different policy responses, examples of which are listed below:
Additional bandwidth allocation and load balancing for a Video Streaming application that requires high bandwidth and low latency.
Traffic prioritization and content compression for an augmented reality application that requires high bandwidth, high data speeds and low latency.
Traffic prioritization and edge caching for a remote surgery application which requires ultra-reliable low-latency.
At the network level, different traffic conditions invoke different traffic routing and filtering processes, effectively eliminating an adverse network event even before it takes place.
Delivering deeper insights
While traffic management is pushing DPI deployments across today’s networks, be it as an appliance or a virtualized function on a NFV infrastructure (NFVi), DPI’s contribution to network analytics is expected to see another wave of adoption, especially across service provider and enterprise networks. Network analytics uses DPI’s traffic classification data and merges this with data sources from outside the network to create detailed and meaningful insights. These insights can range from application trends (usage of selected applications by type of customers) to data consumption trends (usage of data by time and type of mobile data plans) and device trends (usage by type of end devices). Correlation analysis coupled with long-term trend analysis creates a deep understanding of network behaviour, customer consumption patterns and implication of applications types on both costs and revenues of an enterprise or service provider. All these data feeds into enterprise decision making at various levels.
Powering AI-driven networks
Analytics in recent years has taken a new turn with the incorporation of machine learning (ML) and artificial intelligence (AI). Huge data sets generated by service providers and enterprises within their applications and their networks are being processed not just to create meaningful insights. The recurring data patterns are identified by the system and are automatically incorporated into real-time decision making. ML and AI are transforming analytics from a tool that reads the past to a tool that predicts the future. DPI which provides the essential packet-level inputs for network analytics will thus become a critical element in AI-driven networks - fuelling the information necessary for AI-driven decision making and predictive maintenance at both an application- and network-level.
Tackling network security
A key aspect of intelligent networks is the speed of the response to security threats. With the advancement in IP communications come threats in all forms. Ransomware, Trojan, virus and distributed denial of service (DDoS) attacks continue to proliferate on the web, and newer, more advanced forms of cyberattacks are registered every day across enterprise and service provider applications and networks. Failure to tackle network security can bring today’s digital-driven businesses to a halt. DPI’s ability to identify and classify threats including zero-day exploits in real-time is well recognized and is widely documented. Today, these capabilities have grown to cover both encrypted and obfuscated traffic via a vast library of threats updated in specialist labs, at rapidly increasing frequencies. Via DPI, suspicious traffic is identified in real-time and this information is fed to network services such as firewalls and intrusion prevention systems such that threats are blocked and eliminated before they hit the next network node.
Driving monetization
With the deployment of intelligent networks, network awareness goes beyond supporting traffic management decisions, analytics and security policies. It is seen as a new asset for network owners as it helps them not just design their future networks but also refine their monetization strategies. Mobile networks in the 5G era for example, feature network slicing, which requires traffic from different end-applications to be routed through different logical networks based on the same hardware layer. Sizing a slice, allocating the bandwidth within that slice, and pricing that slice requires packet level insights on the application and application attribute (for example, transfer of video images within a smart city application) matched against overall network resources comprising not just the bandwidth but also computing and storage resources consumed along the application delivery pathway. Towards this end, DPI is expected to drive monetization by providing product teams across the likes of mobile operators, WAN providers and fixed network operators application-level intelligence that allows them to develop flexible, usage-based pricing options for their customers.
While the scope of DPI functionalities - from traffic analytics to QoS optimization - has been increasing consistently over the last 20 years, recent years have witnessed DPI growing into a much more critical capability within the network. Against a backdrop of traffic that continues to grow phenomenally, and the growing breadth of policies and services needed to keep networks running, DPI lends intelligent networks of today what they need most - intelligence.
Learn more about DPI and how it delivers network awareness via Rohde and Schwarz’ latest whitepaper ‘DPI – A Crucial Enabler for Network Awareness’. Download your free copy today.
Also join Rohde & Schwarz at MWC Los Angeles at booth 2024 in hall south on October 22-24 to see a demo of a market-leading DPI software.
This is a sponsored article.
