Peering at CNX

The Internet: A Network of Networks

The Internet is a federation of more than 65,000 independently operated networks, known as Autonomous Systems (ASNs).

Each Autonomous System:

  • Operates its own IP address space,
  • Defines its own routing policy,
  • Decides with whom it interconnects.

Networks interconnect using the Border Gateway Protocol (BGP). Through BGP, networks exchange reachability information describing:

  • Which IP prefixes they originate,
  • Which prefixes they can reach,
  • The AS paths associated with those prefixes.

This exchange of routing information allows each network to build its routing table and determine how traffic should be forwarded.

All interconnection on the Internet is based on this mechanism. When two networks exchange only their own prefixes, the relationship is commonly referred to as peering. When a network agrees to provide access to a much larger set of external prefixes, the relationship is commonly referred to as transit. Technically, both models rely on the same BGP exchange of reachability information; the distinction lies in routing policy and scope.

Domestic Interconnection and Local Reachability

CNX brings together networks that operate in Cambodia, including:

  • Internet Service Providers (ISPs),
  • Mobile network operators,
  • Financial institutions,
  • Government networks,
  • Content platforms,
  • Cloud and infrastructure providers,
  • Enterprise networks.

When these networks interconnect at CNX and exchange routes, traffic between them flows directly across the exchange fabric. Communication between Cambodian networks can therefore occur entirely within the domestic interconnection environment.

If a service is hosted in Cambodia and accessed by users in Cambodia, and both networks participate at CNX, routing information is exchanged locally and packets are delivered directly between those networks.

This strengthens:

  • Domestic routing resilience,
  • Latency performance for local services,
  • Operational independence of national infrastructure,
  • Reduced exposure to international transport disruptions.

Global connectivity remains achievable through external interconnections where necessary. However, for communication between Cambodian networks, CNX enables direct domestic interconnection.

The Role of CNX

The Cambodia Network Exchange (CNX) is a Layer-2 Internet Exchange Point (IXP).

Its primary function is to provide shared Ethernet switching infrastructure that enables independent networks to interconnect directly within Cambodia. Participating networks establish physical connections to the exchange and exchange routing information using BGP.

CNX provides:

  • Layer-2 interconnection across a shared peering fabric,
  • A BGP route server service to facilitate multilateral route exchange,
  • Route validation services supporting IRR and RPKI origin validation,
  • Supporting infrastructure services for operational stability, including time synchronization and diagnostic tooling.

The Layer-2 exchange fabric carries traffic directly between member networks. The BGP route server infrastructure operates in the control plane, distributing validated routing information among participants.

These components are logically separated:

  • The exchange fabric provides connectivity.
  • The route servers provide control-plane coordination.
  • Each participant independently determines routing policy and forwarding behavior within its own network.

CNX operates as neutral interconnection infrastructure. It enables networks to exchange routes and traffic directly while preserving full routing autonomy for each participant.

Control plane and data plane separation at CNX

Traffic flows directly between participant networks across the Layer-2 fabric. BGP route servers operate in the control plane.

Route Servers at CNX

In a shared exchange environment, each participant could establish individual BGP sessions with every other participant. As the number of networks grows, this becomes operationally complex. To simplify multilateral interconnection, CNX operates route servers.

Bilateral vs Route Server peering model

Left: Bilateral peering model. Right: Multilateral peering via CNX route servers.

A route server:

  • Receives BGP route announcements from participants,
  • Redistributes those announcements according to defined policy,
  • Does not forward traffic,
  • Does not insert itself into the forwarding path.

The route server reduces the number of bilateral BGP sessions required, while preserving direct traffic exchange between networks. Peering via the route server is mandatory at CNX.

What Changes When You Connect to CNX

Connecting to CNX introduces additional external BGP adjacencies and therefore additional potential ingress and egress paths.

In environments with a single upstream provider, traffic may predominantly enter and exit through one edge router. After connecting to CNX:

  • Multiple networks may send traffic to you directly,
  • Inbound traffic may arrive via different edge routers,
  • Path selection is determined by standard BGP decision processes across independent networks,
  • Traffic symmetry cannot be assumed.

This is a normal characteristic of multi-homed Internet connectivity. From an architectural perspective, connecting to CNX transitions your edge from a primarily upstream-based model to a distributed interconnection model. Internal routing, firewall policies, and stateful services should be designed accordingly.

Implementation Overview

Connecting to CNX involves:

  1. Provisioning a physical cross-connect to a CNX facility.
  2. Configuring Layer-2 connectivity to the assigned peering VLAN.
  3. Establishing BGP sessions with the CNX route servers.
  4. Ensuring IP prefixes and downstream ASNs are correctly registered in IRR and covered by valid ROA records.
  5. Applying appropriate inbound and outbound routing policies.
  6. Verifying route exchange and traffic behavior.

Detailed technical documentation is available in the following sections: