Internet Connections - Part 2
A business has needs a home never has: guaranteed bandwidth, contracted reliability, and a way to link several sites securely. This lesson covers the enterprise WAN toolbox — Metro Ethernet, leased lines, MPLS, VPN over broadband, and cellular backup — and the service-level agreement that makes a provider accountable for what it sells.
What a business WAN must provide
A wide area network (WAN) links locations that are too far apart for you to own the cabling yourself, so a business leases capacity from one or more providers. Unlike the home connection of Lesson 5, a business WAN is usually chosen for four things beyond raw speed.
Committed bandwidth
A guaranteed rate, not a best-effort share of someone else’s cable.
Predictable performance
Bounded latency, jitter, and packet loss for real-time applications.
Contracted reliability
An availability figure the provider must meet — the SLA.
Secure site-to-site connectivity
So branch, head-office, and cloud traffic stays private.
The provider cloud hides the internal paths. From the customer’s point of view there is exactly one hand-off (the CE–PE link) at each site, and the provider is responsible for moving traffic between them.
The enterprise WAN services
Five service families cover almost every business WAN you will meet. Pick one to see its speed, cost, and reliability profile.
Metro Ethernet
Metro Ethernet extends Ethernet out of the building and across a provider’s metropolitan or regional network, usually over fibre. Because the customer hand-off is the same Ethernet already used inside the LAN, there is no protocol conversion and no expensive serial equipment to buy. Providers sell it in two common forms: E-Line (a point-to-point virtual circuit between two sites) and E-LAN (a multipoint service connecting several sites as one logical Layer 2 network). It scales smoothly by increasing the contracted speed.
Dedicated leased lines (T1/T3, E1/E3)
A leased line is a dedicated point-to-point circuit reserved for one customer, with fixed symmetric bandwidth in both directions. This is the classic definition of dedicated bandwidth from Lesson 5. The traditional digital carrier rates are worth knowing because the exam still references them.
| Circuit | Capacity | Built from | Region |
|---|---|---|---|
| T1 | 1.544 Mbps | 24 channels of 64 kbps (DS0) | North America, Japan |
| T3 | 44.736 Mbps | 28 T1 circuits | North America, Japan |
| E1 | 2.048 Mbps | 32 channels of 64 kbps | Europe and most of the world |
| E3 | 34.368 Mbps | 16 E1 circuits | Europe and most of the world |
Leased lines are extremely reliable and give predictable latency, but they are expensive and inflexible compared with Ethernet or MPLS services, and many providers are retiring them. They remain the clearest example of a dedicated, symmetric circuit.
Broadband plus VPN, and cellular
Not every branch justifies a leased line or an MPLS port. A small branch can use an ordinary broadband connection (cable, DSL, or fibre) — cheap and widely available, but shared and best-effort with no committed bandwidth. To make it secure and private, the branch builds a VPN tunnel across the public Internet back to head office. The two common forms are the site-to-site VPN (branch router to HQ router) and the remote-access VPN (an individual user connecting in). The tunnel commonly uses IPsec to provide confidentiality and integrity. The trade-off is clear: low cost and easy setup, but performance depends on the public Internet rather than a provider guarantee.
Cellular is rarely the primary WAN for a large site, but it is excellent as a backup link. A router with both a wired circuit and a cellular modem can fail over automatically to 4G/5G if the wired link drops. Because it needs no cable, cellular is also useful for temporary sites, pop-up locations, and disaster recovery. Its unpredictability and metered data make it secondary rather than primary for most businesses.
| Option | Speed | Cost | Reliability / SLA | Best for |
|---|---|---|---|---|
| Metro Ethernet | High, scalable | Medium | High; SLA available | Scale Ethernet between nearby sites |
| Leased line (T1/T3, E1/E3) | Fixed, symmetric (low by today’s standards) | High | Very high; SLA | Dedicated predictable point-to-point circuits |
| MPLS | Medium–high | Medium–high | High; QoS and SLA | Multi-site QoS-sensitive traffic |
| Broadband + VPN | Varies (shared) | Low | Best-effort; no SLA | Small branches and remote workers |
| Cellular | Varies (4G/5G) | Medium (metered) | Variable | Backup, temporary, and mobile sites |
Inside MPLS
Multiprotocol Label Switching (MPLS) is a provider WAN service defined in RFC 3031. Instead of routing every packet by looking up the IP destination at each hop, an MPLS network attaches a short label to each packet at the provider edge and forwards it along a pre-established label-switched path (LSP). Because forwarding is based on the label, the provider can engineer traffic paths and apply quality of service precisely — which is why MPLS is valued for voice, video, and other delay-sensitive traffic.
- Customer edge (CE) — the customer’s router at each site.
- Provider edge (PE) — the provider router the customer connects to.
- Provider (P) — provider core routers that switch labels internally.
MPLS is “multiprotocol” because it can carry IP and other traffic, and it is commonly sold as a managed MPLS VPN that keeps each customer’s traffic logically separate while still giving strong QoS and an SLA. The customer sees only the CE–PE hand-off at each site — every label, LSP, and core path between them belongs to the provider.
Service-level agreements
An service-level agreement (SLA) is a contract between a customer and a provider that states measurable guarantees for the service, and the remedies if those guarantees are missed. It is what separates a business-grade WAN from a consumer broadband line: the consumer contract makes no uptime or latency promise, while the business contract does.
| Metric | What it measures |
|---|---|
| Availability / uptime | Percentage of time the service must be usable (e.g. 99.9%). |
| Throughput / bandwidth | The committed data rate the provider must deliver. |
| Latency | Maximum round-trip delay allowed, critical for real-time traffic. |
| Jitter | Variation in delay between packets, which disrupts voice and video. |
| Packet loss | Maximum percentage of packets dropped. |
| MTTR | Mean time to repair after a fault. |
ISP tiers, colocation and data-centre interconnect
Business WANs ultimately ride on the Internet’s provider hierarchy (introduced in Week 1, Lesson 1).
- Tier 3 — local ISPs that sell directly to homes and small businesses.
- Tier 2 — regional and national providers that buy transit from Tier 1 and sell to Tier 3.
- Tier 1 — global backbone carriers that interconnect with each other settlement-free (no charge for peering) and can reach the entire Internet without paying for transit.
Colocation means renting space for your own servers inside a provider’s data centre: you get power, cooling, physical security, and network connectivity, while you still own and manage the equipment. Colocation matters because data centres act as the meeting points of the Internet. Carriers cross-connect inside the building, and providers exchange traffic at Internet exchange points (IXPs), typically in a meet-me room. Placing your servers in a colo puts you physically next to many carriers and cloud on-ramps, which shortens paths and lowers the latency for the services you depend on.
Resiliency: dual links and failover
A business WAN with a single link has a single point of failure: when that link breaks, the site is offline. Resiliency is engineered in at several levels, and you can test whether a design survives each failure mode.
- Dual links — two connections (for example MPLS as primary and broadband VPN or cellular as backup).
- Diverse paths — the two links physically enter the building by different routes, so one cable cut does not take out both.
- Failover — automatic switch to the backup when the primary fails.
- Redundant hardware — duplicate routers, switches, and power supplies.
- Multi-homing — connecting to more than one ISP so a single provider outage does not disconnect the site.
This is the WAN expression of the fault-tolerance pillar you will study in the next lesson: no single failure should take the whole network down.
Key terms
| Term | Meaning |
|---|---|
| WAN | Wide area network; links sites too far apart to cable yourself, leased from a provider. |
| Metro Ethernet | A provider service extending Ethernet (E-Line or E-LAN) across a metro or regional network. |
| Leased line | A dedicated point-to-point circuit with committed symmetric bandwidth. |
| MPLS | Multiprotocol Label Switching; label-based provider WAN service (RFC 3031) with QoS. |
| LSP | Label-switched path; the predefined route an MPLS label follows. |
| CE / PE / P | Customer edge, provider edge, and provider core routers in an MPLS network. |
| VPN | Virtual private network; a private tunnel (often IPsec) over a shared network. |
| SLA | Service-level agreement; contracted performance guarantees with remedies. |
| Colocation | Renting space and power for your equipment in a provider data centre. |
| Failover | Automatic switching to a backup link or device when the primary fails. |
Review
Check what you have learned with these review questions.
Sources & references
- Rosen, E., Viswanathan, A. & Callon, R. RFC 3031: Multiprotocol Label Switching Architecture (2001). rfc-editor.org/rfc/rfc3031.
- Cisco. MPLS: Multiprotocol Label Switching (technology documentation). cisco.com.
- Cisco. CCNA (200-301) Exam Topics, Network Fundamentals (WAN technologies). cisco.com.
- Kent, S. & Seo, K. RFC 4301: Security Architecture for the Internet Protocol (IPsec) (2005). rfc-editor.org/rfc/rfc4301.
Target exams: CCNA 200-301 v1.1 · CyberOps Associate 200-201. Last content verification: 2026-10-11.