Business fiber & dedicated internet

The medium and the service are separate purchases

Fiber describes the path; the service sold over it is a separate purchase. A fiber-to-the-premises service delivered over a shared passive optical network is still shared, and a provider can sell a shared service or a committed one over the same strand. The word that matters in an offer is not fiber but the commitment attached to it. Dedicated internet access is a commercial description rather than a physical guarantee of exclusive fiber: what it should mean is a provisioned circuit with a committed information rate, stated service commitments and defined remedies. You verify that in the contract, not in the product's name, and two offers that both say dedicated can differ in every one of those clauses.

• Fiber is the path; the service definition is the product.
• Dedicated describes a commitment, not necessarily exclusive fiber strands.
• Ask for the committed information rate, the symmetry and the availability target.
• Verify each of those in the contract, not the marketing.

The handoff: what actually arrives at your premises

The access network terminates at an optical network terminal, the customer-premises equipment that presents an interface — usually Ethernet — to your own network. That terminal is not by itself proof that the service is dedicated or carries any particular commitment; it is simply the conversion between the provider's fiber and your copper. The boundary of responsibility, the demarcation point, is fixed by the service agreement and the installation design, and it may sit at the terminal, at a smart jack or at another handoff, so do not assume the equipment's location and the contractual demarcation are the same place. Write down who owns, powers, monitors, replaces and troubleshoots the equipment on each side of that line, because the sentence you rely on at two in the morning is the one you agreed before the install.

• The optical network terminal converts the provider's fiber to a customer-facing interface.
• The contractual demarcation may not sit where the equipment does; confirm it in writing.
• Fix who owns, powers, monitors and replaces each piece of equipment.
• Know what interface is presented and what cabling reaches your own kit.

Carrier Ethernet is a defined service, not a marketing tier

Above the physical layer, a fiber service is usually described in a vocabulary that has published definitions. MEF, the industry body that writes the Carrier Ethernet service specifications, defines an Ethernet Line, or E-Line, as an Ethernet service based on a point-to-point Ethernet Virtual Connection; the document is MEF 6.3, Subscriber Ethernet Services Definitions. That matters because an E-Line is a point-to-point service type — not a synonym for internet access, and not a promise of a physically exclusive fiber pair. Bandwidth, service multiplexing, class of service and performance objectives all depend on the specific offer, and a Carrier Ethernet service can be delivered over shared provider infrastructure. At the access layer the standard is Ethernet in the First Mile, approved as IEEE Std 802.3ah-2004, which standardises Ethernet over access media rather than defining a commercial product. The value of naming these is that a buyer can quote them back and ask which definition an offer is written against.

• MEF 6.3 defines the subscriber Ethernet service types, including E-Line over a point-to-point EVC.
• An E-Line is a layer-2 service, not internet access, and not necessarily exclusive fiber.
• IEEE Std 802.3ah-2004 (Ethernet in the First Mile) standardises the access medium, not the product.
• Ask which published definition the offer is written against.

Lit fiber and dark fiber are different instruments

Two arrangements travel over the same glass and they are not interchangeable. With lit fiber the provider supplies the transmission equipment and sells a service over the fiber, so you buy the outcome and the provider owns the link. With dark fiber you lease unlit strands and supply your own optical equipment, so you also take on the link design and its maintenance. A dark-fiber lease should identify the strand count, the route, the access and splice rights, the repair obligations and an acceptance criterion for optical loss, because none of those come with the strand by default. Dark fiber makes sense for a business connecting two sites it controls, or one with the engineering capacity to run its own optics; most businesses are better served by a lit service with a written commitment.

• Lit fiber: the provider supplies the electronics and the service.
• Dark fiber: you lease strands and supply the optics and the link design.
• A dark-fiber lease should state strand count, route, splice rights, repair duty and loss criteria.
• Choose by whether you can run the optics yourself.

Installation is a project with dependencies

A circuit order is often a construction project wearing a service label. Where fiber has to reach the building, the work can include extending the route, gaining building access or rights-of-way, and rearranging existing attachments on shared poles. In the United States the pole work is scheduled under the FCC's rules at 47 CFR 1.1411, which set out the sequence: the utility surveys, presents an itemised estimate of the necessary make-ready, then completes the make-ready on a clock that is longer for larger orders. The rule distinguishes mid-sized and large orders by pole count and gives each its own timetable, and it lets the new attacher elect to do the work if the utility or an existing attacher says it cannot meet the deadline. Ask for a written schedule with the survey, build and permit dependencies named against dates, and make the delivery date contingent on the third-party work rather than on the provider's optimism.

• Getting fiber in can require a route extension, building access and rights-of-way.
• Pole work runs on the make-ready sequence and clocks in 47 CFR 1.1411.
• The rule scales the timetable by order size, measured in poles.
• Ask for a written schedule with each dependency named against a date.

The contract terms that decide what you bought

Four clauses carry most of the practical value. The service definition fixes the committed information rate, the symmetry and the class of service. The service level section states availability, latency, packet loss and repair commitments, how each is measured over what window, and the remedy — usually a credit against the charge. The term section fixes the length, the notice period and whether construction or equipment costs are recovered if you leave early. The addressing section says how many usable addresses you receive, who owns them and whether reverse DNS is delegated to your control. Read them together, because each one changes the meaning of the others: a generous availability commitment on a short term is worth less than a modest one you can rely on for as long as you expect to stay at the site.

• Service definition: committed rate, symmetry and class of service.
• Service level: what is measured, over what window, and the credit.
• Term: length, notice period and any construction-cost recovery.
• Addressing: block size, ownership and reverse DNS delegation.

Addressing, routing and what is hard to change later

If anything your business runs has to be reachable from outside — a mail server, a remote-access gateway, a hosted phone system, a site-to-site link — then the addresses matter, and they usually belong to the provider for the life of the service. Moving them means changing every place they appear, which is why the addressing clause belongs in the first reading of the contract rather than the last. If you intend to run your own routing between providers, or to fail over between two circuits, that is a different tier of service: it needs the provider to support dynamic routing, and in practice that means exterior routing — the protocol a network uses to advertise reachability to other networks, standardised as the Border Gateway Protocol in RFC 4271. Confirm in writing whether dynamic routing is available at the tier you are buying, and whether the provider will accept your own autonomous-system number.

• Static and publicly reachable addresses usually stay with the provider.
• Ask for the block size, ownership and reverse DNS delegation in writing.
• Dynamic routing between circuits is a higher service tier; confirm it is offered.
• BGP is the standard exterior routing protocol (RFC 4271); the provider must support it.

What to send with a request for pricing

The file a supplier needs is the same one that makes two offers comparable. Bring the full service address including suite or unit, the measured performance of what serves the site now rather than its advertised figure, what the business uploads as well as downloads, the applications that would be noticed immediately if they degraded, any requirement for publicly reachable addresses or dynamic routing, who controls building access, the current service's contract end date and notice period, and the deadline by which any change must be complete. If the premises is shared or multi-tenant, say so, because that changes who has to be involved in access and riser decisions. Everything else follows from those facts, and a supplier who has them can quote against your requirements instead of a template.

• Full address, suite detail and who controls building access.
• Current measured performance, both directions.
• Applications with latency sensitivity, and any addressing or routing requirement.
• Existing contract end date, notice period and the deadline for change.

PUT THE OPTIONS SIDE BY SIDE

Your comparison checklist

Common questions