Lease line internet (LLI), also sold as dedicated internet, is a fibre connection where the bandwidth you pay for belongs to you alone. Nobody in the next building slows you down at 9 p.m. You get the same speed up and down, a block of real public IP addresses, and a service level agreement (SLA) that puts the provider's repair time in writing.

On the networks I provision, LLI is a routed (Layer 3) service. Each customer gets their own VLAN, a /30 point-to-point link and a /28 public block. The fibre leaves a ZTE or Huawei switch on a single fibre core and ends at the customer in a media converter or an SFP port.

LLI or shared broadband

The difference is contention. Broadband shares capacity between many users and promises a "best effort" speed. LLI reserves the full rate for one customer.

Why businesses pay for it

Three things push a customer to LLI: guaranteed upload, public IPs, and an SLA.

Offices that host their own servers, run site-to-site VPNs or carry voice over IP need all three. A video call that freezes because a neighbour started a download is a broadband problem. It does not happen on a reserved line.

Then there are the short jobs. For temporary events and live broadcasting, we bring the fibre to the venue, put a media converter at the end, and plug it straight into the broadcaster's laptop. No router, no LAN. The laptop takes the customer address on the /30 and goes live.

How the connection reaches you

Diagram of a lease line path: MPLS backbone, core router, aggregation switch, access switch, single-core fibre, media converter, then customer router, switch or laptop

Traffic follows the same path every time, in both directions:

1. Backbone. The provider's core carries traffic on MPLS, and routes are exchanged with BGP.
2. Core router. Routes the customer's /28 block towards them and announces it upstream.
3. Aggregation switch. Collects the VLANs from many access switches and hands them up to the core.
4. Access switch. The ZTE or Huawei switch nearest the customer. The customer's fibre lands on one of its SFP ports.
5. Fibre. A single fibre core carries both directions. Reach depends on the SFP fitted at each end.
6. Media converter. Turns light back into copper Ethernet at the customer's end.
7. Customer router, switch or laptop. Uses a static default route to the provider's end of the /30.

VLANs, IP addressing and bandwidth

One VLAN per customer#

Every LLI customer gets their own VLAN on the provider's switches. It keeps each customer's traffic apart from everyone else's. It also gives the provider one clean place to attach that customer's addressing and bandwidth limit.

A /30 link and a /28 public block#

The /30 is the handover link: four addresses, two of them usable. One sits on the provider's side, one on the customer's router, switch or laptop. The /28 is the customer's public block, routed to their side of the /30.

This is what a customer receives at handover. The values below are from a documentation range, not a live circuit:

IP:       203.0.113.218
Subnet:   255.255.255.252
Gateway:  203.0.113.217
PoP:      <PoP switch name>
Port:     gei-0/x/x/x

The subnet mask 255.255.255.252 is the /30. In this /30, 203.0.113.216 is the network address and 203.0.113.219 is broadcast. The provider's end of the link takes .217 and the customer takes .218. The port line is the ZTE interface name on the PoP switch, which is what the field team needs when a fault is reported.

The /28 needs a customer router to use it. Put it on the router's LAN side and 14 addresses are left for hosts, one of which the router takes. That leaves 13 for servers and devices. A laptop or a plain switch on the line uses only the /30 address.

Routing and the backbone#

Inside the provider network, customer traffic crosses an MPLS backbone and BGP carries the routes, including the customer's /28. The customer never sees any of this. From their side it is one link and one gateway.

Bandwidth is limited, not shared#

The provider caps each customer's VLAN at the contracted rate. Buy 50 Mbit/s and you get 50 Mbit/s, in the morning and at peak hours, because nobody else's traffic sits inside your limit.

The components#

Core router#

The top of the provider's own network. It runs MPLS and BGP and announces customers' public blocks upstream.

Aggregation switch#

Sits between the core and the access layer. One aggregation switch collects the uplinks from many access switches, so the core needs fewer ports.

Access switch#

The last provider switch before the customer, ZTE or Huawei in my work. Each LLI customer gets a fibre port with an SFP and a VLAN on it.

SFP module#

The small pluggable optic that turns the switch's electrical signal into light. Our links run on a single fibre core, so each SFP sends on one wavelength and receives on another. Reach depends on the module; longer links take longer-reach SFPs.

Fibre and patch cords#

One fibre core runs from the provider's optical distribution frame (ODF) to the customer. Patch cords with LC or SC connectors join it to the SFP at the switch and to the media converter at the customer.

Media converter#

A box with a fibre port on one side and an RJ45 Ethernet port on the other. It lets a router, switch or laptop with no SFP slot use the fibre. The ones I install are D-Link.

Customer router, switch or laptop#

Whatever plugs into the media converter takes the customer end of the /30 and a static default route to the provider's gateway. In an office that is a router, which also holds the /28 on its LAN side, NAT and the firewall. Some customers plug in a switch. For event jobs it is a laptop.

The numbers that matter#

What it costs#

LLI is billed per Mbit/s per month, with VAT on top. A connection fee and a local loop charge come on top of that. The rate per Mbit/s usually falls as the bandwidth grows, so ask your provider for its current tariff before you size the link.

Disadvantages#

LLI costs more than broadband for the same headline speed. You pay for capacity nobody else uses.

It takes longer to install. If no fibre reaches your building, the provider has to lay it first.

One fibre is one point of failure. The service runs on a single core with no second path, so a cut on the local loop takes it down until the fibre is spliced.

And a public IP is exposed. A laptop plugged straight into the media converter sits on the open internet with no router in front of it. Turn its firewall on before the event starts.