Service provider network architecture: access, aggregation, core, peering and transit
In short: A service provider network is layered like a campus but at national scale. Access networks reach homes and businesses, aggregation collects that traffic, and a high-capacity core moves it between cities. At the edge the provider swaps traffic with other networks through peering and paid transit, often at internet exchange points, using BGP. Content caches inside the network keep popular traffic local.
Why enterprise engineers should understand provider networks
Almost every enterprise depends on someone else's network: the broadband line at a branch, the leased line at head office, the MPLS service between sites, and the internet path to every cloud application. When a branch is slow, the cause is often in the provider's network. Knowing how that network is built helps you read a traceroute, choose the right service and ask the provider sensible questions.
The layers of a provider network
- Customer premises equipment (CPE): the router or optical terminal at the customer's site. It may be owned by the provider or the customer.
- Access: the "last mile" technology that reaches the customer: fibre (often passive optical networks), cable, copper DSL, fixed wireless or mobile 4G and 5G.
- Aggregation: equipment in local exchanges that collects thousands of access connections and hands them upwards. Broadband sessions are usually authenticated and given addresses here.
- Core: a small number of very fast routers in major cities joined by high-capacity fibre. The core moves traffic, it doesn't inspect it. Many providers run MPLS or segment routing here so they can steer traffic and offer private services.
- Points of presence (PoPs): buildings where the provider has equipment and where customers or other networks connect.
Mobile operators follow the same pattern: radio sites form the access layer, backhaul links aggregate them, and a mobile core handles subscribers. See 5G architecture for the mobile side.
Peering, transit and exchange points
No single provider reaches every network on the internet, so providers connect to each other. They exchange routes using the Border Gateway Protocol (BGP), where each network is identified by an autonomous system number.
| Arrangement | What it means | Who pays |
|---|---|---|
| Transit | A larger provider carries your traffic to the whole internet | The customer pays the transit provider |
| Peering | Two networks swap traffic only for their own customers | Usually free, as both sides benefit |
| Internet exchange point (IXP) | A shared switching fabric where many networks peer in one place | Members pay for the port and membership |
Large content and cloud companies also place caches inside provider networks, so a popular video is served from a nearby city rather than another continent. This is why traffic to big platforms often leaves your provider's network after very few hops.
Services a provider sells to businesses
- Dedicated internet access: an uncontended line with a service level agreement (SLA) for availability and repair time.
- Business broadband: cheaper, shared capacity, usually with weaker repair targets.
- MPLS VPN or Ethernet services: private connectivity between sites across the provider's core.
- Cloud on-ramps: private connections from the provider's PoP into major cloud platforms.
Tip: two internet lines from the same provider that share the same duct or exchange are not truly diverse. Ask for route diversity in writing, or use a second provider and a different access technology.
Real-world example: a retailer's "resilient" branches
A fictional clothing retailer gave every shop two broadband lines for resilience. When roadworks cut a fibre cable in one town, both lines at three shops went down together, and card payments stopped for a day. The investigation showed both lines ran from the same local exchange through the same duct. The network team switched the backup at each shop to a 4G/5G router from a different mobile operator, added SD-WAN to fail over automatically, and asked their provider for a diversity report for the head office leased lines.
Practice
Put these provider layers in order from the customer outwards: core, access, aggregation.
What is the difference between transit and peering?
Why might two broadband lines from the same provider fail together?
Key takeaways
- Provider networks are layered: customer premises, access, aggregation, core and PoPs.
- Providers exchange routes with BGP through paid transit and peering.
- Internet exchange points let many networks peer in one place.
- Content caches inside provider networks keep popular traffic local.
- Real resilience needs diverse physical paths, not just two lines.
Original content written by Mohamed Fayas for FM TecHub. © 2026 FM TecHub. You're welcome to share a link to this page; please don't copy the text elsewhere without permission.