Many networks connect to outside providers to fulfill their traffic needs. These connections are often referred to as peering sessions, and the locations where they occur are known as Internet exchange points (IXPs).
In today’s world, Internet exchange points (IXPs) are an integral part of the Internet ecosystem and, in many countries, are defined as critical infrastructure. IXPs provide interconnects for various Internet service providers (ISPs) to exchange traffic locally, resulting in faster Internet connectivity. In the past, when transit-and transport-network prices were high, peering provided a way to lower end-user costs. Likewise, the increased number of paths available through the IXP improves routing efficiency and fault tolerance.
The Internet exchange points play a vital role in the Internet ecosystem as there are over 400 IXPs in over 100 countries today worldwide. The largest IXPs carry peak data rates of around 10 Tbps and connect hundreds of networks.
And with the advent of 5G, the first generation of cloud-native mobile technology, IXPs are evolving. 5G’s service-driven architecture includes RAN, 5GC, hyperscalers, and transport. Transport is now becoming a key part of 5G infrastructure.
Therefore, IXPs are transitioning from standard legacy Ethernet and IP networks such as Layer 2 virtual local area networks (VLANs) and virtual private LAN services (VPLS) into data center architectures, with leaf-spine switching fabrics. For examples, Clos architectures utilizing Ethernet VPN-Virtual Extensible LAN (EVPN-VXLAN) with automation and orchestration.
Additionally, part of this evolution also includes Data Center Interconnect (DCI) connectivity. DCI backhaul and traffic transport provide more value-added services, such as multi-access edge computing (MEC), neutral host networks (NHNs), and private network interconnects (PNIs). Keeping all this in mind, the IXPs of yesterday are facing a challenge as the evolution drives them towards becoming more of interconnects than peering points only. The interconnects will become an integral part of 5G infrastructure to reduce total cost of ownership (TCO), increase automation, provide software-defined infrastructure, and facilitate new value-added services in addition to basic peering services, as shown in Figure 1.
Figure 1. Internet Exchange Points evolution into Interconnect 2.0
The Era of Interconnect 2.0 is Here…
We are entering a new era of Interconnect 2.0, providing value-added services where technological developments such as 5G and Edge Compute will enable IXPs, as well as governments, enterprises, and consumers to coordinate their activities more. This era will lead to faster coordination and more efficient use of various resources.
The critical component of the new Interconnect 2.0 will become the universal availability of flexible and fast, high-reliable, low-latency networks providing transport for a myriad of applications. This foundation will enable a complex array of communications, decisions, transactions, and processes to be completed quickly and, in many cases. automatically without human intervention, using machine learning (ML) and artificial intelligence (AI). With 5G, interconnects will require automated network slices for 5G traffic with various stringent service level agreements (SLAs). The network, or in other words, the interconnect, becomes key.
Historically, telecom and IXP networks have been created using dedicated hardware and software to achieve a specific function or serve a specific use case. This has ensured that these networks are reliable and somewhat secure but has also hindered innovation – from not only service providers but also vendors of equipment used in such. As shown in Figure 2, this new era of Interconnect 2.0 will require more from the network than ever before – applications require the network to be flexible, dynamic, and accessible. The network must support a range of technical and commercial options and operations. These networks must impart actionable insights and be flexible and scalable for speed, bandwidth, latency, security, rapid business model changes, and countless other variables to meet the needs of applications and users. The telco cloud, which is becoming the foundation for the cloud-native 5G services, will be the enabler of the future of Interconnect 2.0.
Innovation at Interconnect 2.0: Disaggregation, SDN and Network Automation
Disaggregation, software-defined networking (SDN), and network automation have already been applied and deployed in the telco world for 4G/5G rollouts as data-centric networks. These will become very viable within the IXP networks that will evolve into Interconnect 2.0s. As IXP networks are typically more localized than telco networks and do not need to cope with legacy infrastructure and services, they will be ideal places to introduce new networking concepts.
Disaggregation and openness will hasten innovation. Disaggregation, when applied to the Interconnect 2.0 and transport infrastructure, provides horizontal scalability allowing for unexpected growth without the need to pre-plan large chassis-based system capacity or forklift upgrades.
In disaggregated networks, innovation is efficient, as network functions are decoupled from each other and can run in containers and evolve at their own speed. As a result, interconnects can seamlessly introduce additional steps in interface capacity like 400GbE or 1 Terabit Ethernet, single-chip switching capacities, and functionalities such as programming protocol-independent packet processors [P4].
Realizing the Interconnect 2.0 vision
Moving to Interconnect 2.0 is challenging and will not happen overnight. While the rationale for embracing the new era of interconnects is clear, many factors are slowing the industry’s movement into becoming the interconnects of the future. Despite this, the industry is moving toward cloud-native deployments with tightly integrated ecosystems incorporating APIs for orchestration and automation of services, enabling scalability and thus benefiting the interconnects.