Wavelength Services for Businesses

A wavelength is exactly what it sounds like: one wavelength — one specific color of laser light — on a strand of fiber, carrying your traffic and nothing else. Because modern fiber can carry dozens of different wavelengths simultaneously without them interfering (a technology called dense wavelength-division multiplexing, or DWDM), a carrier can sell you one of those wavelengths as a private, point-to-point circuit between two buildings. You never share it, and nobody else's traffic can congest it.

What makes wavelengths different from almost every other connectivity product is that they're Layer 1: the carrier delivers light, not packets. There's no routing, no IP addressing, no oversubscription, and no carrier equipment inspecting or shaping your traffic. Whatever you put in one end comes out the other — Ethernet frames, Fibre Channel storage traffic, legacy SONET, or a mix of protocols over the same circuit if you multiplex them yourself. To a network engineer, a 100G wavelength between two data centers behaves like a 100-kilometer-long patch cable.

That simplicity is the point. Because the carrier isn't running a routed network between your sites, latency is nearly a pure function of distance (light in fiber travels at roughly 200,000 km/s, about two-thirds the speed of light in a vacuum — call it 5 microseconds per kilometer of fiber route). A wavelength on a 120 km route has a round-trip latency you can calculate, and more importantly, it doesn't change. For applications that care — storage replication, financial transactions, real-time control systems — that determinism is worth real money.

Capacities are standardized in big steps: 1 Gbps and 10 Gbps at the entry level, 100 Gbps as today's mainstream data-center-interconnect workhorse, and 400 Gbps at the top end for carriers, cloud on-ramps, and hyperscale-adjacent workloads. Most business buyers land at 10G or 100G.

One lambda, two endpoints

At each end of the circuit, the carrier hands you a fiber connection — typically a pair of single-mode LC connectors into a transponder or muxponder shelf, or directly into an optic you plug into your own router or switch. Your equipment talks to the carrier's optical transport gear using a standard client interface (10GBASE-LR, 100GBASE-LR4, and similar). From your router's perspective, the far end of the link is the device in the other building, full stop. There is no 'provider network' in between that you need to configure around.

DWDM: many colors on one strand

The reason a carrier can sell you a wavelength without dedicating an entire strand of fiber to you is DWDM. Terminal equipment at each end of a fiber route combines dozens of wavelengths — each a slightly different frequency of light, each carrying an independent signal — onto one strand, and separates them again at the far end. Amplifiers along the route boost all the wavelengths together every 60–100 km or so. Your 100G wavelength rides the same physical fiber as other customers' wavelengths, but optically isolated: congestion on their channels cannot touch yours.

This is also why wavelength pricing scales with route and distance rather than with 'internet speed.' The carrier is reserving optical capacity on a specific physical path between your two buildings, and that path — through specific conduits, huts, and amplifier sites — is what you're paying for.

Protected vs. unprotected paths

A wavelength can be sold unprotected (a single fiber path; a cut between the endpoints takes the circuit down until it's spliced) or protected (the carrier provisions two physically separate paths and the optical gear fails over automatically, typically in under 50 milliseconds). Protected service costs meaningfully more — often 30–60% — because the carrier is reserving capacity on two routes. The critical due-diligence question is whether 'protected' actually means diverse: two wavelengths in the same conduit are one backhoe away from being a single point of failure. Ask for route maps, in writing.

Transparency and protocol independence

Because the service is Layer 1, it doesn't care what you run across it. The same wavelength can carry 100 Gigabit Ethernet today and be repurposed for eight channels of 32G Fibre Channel tomorrow by changing the optics on your side. Jumbo frames, your own VLANs, your own encryption, your own routing protocols — all invisible to the carrier. This is a genuine operational advantage for security-sensitive environments: you can encrypt at Layer 2 with your own MACsec gear and hold the keys, and the carrier never sees a plaintext packet because there are no packets at their layer.

What the carrier manages — and what you do

The carrier is responsible for the optical path: the fiber, amplifiers, and transport shelves, plus optical performance (power levels, bit-error rate) up to the handoff point. You own everything from your router's optic outward: your protocols, your redundancy logic, your monitoring, your encryption. That split is simpler than a managed IP service, but it means a wavelength is only as resilient as the design you wrap around it.

• Replication windows that don't fit: nightly backups and storage replication that can't finish before morning over a shared internet circuit run comfortably inside hours on a 10G or 100G wave.
• Latency variance: routed IP paths can shift route, add hops, or congest at peering points; a wavelength's latency is fixed by geography and stays fixed.
• Security exposure of transit networks: traffic on a wavelength never traverses the public internet or a shared routed backbone — interception means physically tapping the fiber.
• Egress and VPN economics: moving terabytes a month between two facilities through internet circuits or cloud networking often costs more in transfer and throughput fees than a flat-priced optical circuit.
• Capacity ceilings: when a 10G link is at 70% sustained, the honest options are 100G transport or dark fiber — wavelengths get you there without buying fiber strands.
• Vendor sprawl between sites: two internet circuits, two VPN concentrators, and an SD-WAN overlay replaced by one deterministic pipe.

Notice what wavelengths don't solve: they don't give you internet access, they don't connect more than two points, and they don't provide any routing intelligence. A wavelength is a component you design around, not a complete network.

Wavelengths are a specialist tool, and the honest answer for most small businesses is that they're more circuit than you need. If your two sites exchange ordinary business traffic — email, SaaS, some file sharing — SD-WAN over two good internet circuits is simpler and far cheaper. The crossover point is sustained volume, latency sensitivity, or both.

You should be pricing wavelengths when you recognize yourself in one of these profiles: a business with a colocation footprint or private cloud replicating to a DR site; a multi-facility healthcare system moving imaging and records between a primary data center and campuses; a manufacturer linking plants to a central ERP and backup site with real-time MES traffic; a financial or professional firm where milliseconds of jitter have a measurable cost; or a network operator — WISP, MSP, or regional carrier — buying wholesale capacity to feed towers, POPs, or customer tails.

The technical tell is in your monitoring: sustained multi-gigabit flows between the same two endpoints, day after day. Once that flow is a permanent feature of your architecture rather than a spike, flat-rate optical transport almost always beats metered or contended alternatives.

1. Data center interconnect (DCI): linking a primary data center or colocation cage to a secondary site for replication, clustering, and stretched storage networks — the classic wavelength workload.
2. Backup and disaster recovery: dedicated high-capacity pipe for nightly full backups and continuous replication, so recovery point objectives are measured in minutes instead of hours.
3. Campus and metro interconnect: connecting headquarters, plants, or hospital campuses across a metro area at 10G–100G without building a routed WAN.
4. Cloud and carrier hotel access: high-capacity transport into a carrier-neutral facility where cloud on-ramps and carriers meet, instead of hauling traffic over the internet.
5. Wholesale backhaul: WISPs and MSPs leasing 10G/100G waves to feed towers, aggregation sites, and downstream customers.
6. Latency-critical links: trading floors, research computing, and real-time industrial control where the shortest possible fiber route is a requirement, not a preference.
7. Storage-area network extension: Fibre Channel over wavelength between sites for synchronous storage replication.

Wavelength pricing is quote-driven — every circuit is priced on its specific route — so treat any published number as directional at best. What actually drives the monthly recurring charge:

• Capacity: the jump from 10G to 100G is typically far less than 10× in price — 100G is often the better value per gigabit where both are available.
• Distance and route: metro circuits (same market) cost dramatically less than regional or long-haul; long routes traverse more amplifiers and more carrier infrastructure.
• Protection: unprotected service is the base price; protected (diverse-path, auto-failover) service typically adds 30–60%.
• Building readiness: if both endpoints are already 'lit' (carrier equipment present, fiber in the building), pricing is sharp. A near-net building may add construction fees; an off-net building can add thousands in non-recurring construction cost — or make a carrier non-viable entirely.
• Term: one-year terms exist but three years is the pricing sweet spot; longer terms lower the monthly rate and can amortize construction.
• Cross-connects: in data centers and carrier hotels, the facility charges separately for the physical cross-connect between the carrier's cage and yours — a monthly fee per connection that buyers routinely forget to budget.
• Equipment and optics: some carriers include the transponder and handoff optics; others charge. On your side, a 100G optic for your router is a real line item.

As a rough orientation only — not a quote — unprotected 10G metro waves commonly price in the low-to-mid hundreds of dollars per month range, with 100G metro in the low thousands, varying widely by market, route, carrier, and term. Long-haul and protected service run higher. The honest comparison is total three-year cost: monthly recurring plus construction, cross-connects, and your own optics.

A wavelength order has more engineering in it than a broadband install, and the process reflects that:

1. Requirements: endpoints (exact addresses or facility meet-me rooms), capacity, protection level, latency target, and handoff interface. Get the optics spec right here — it determines what you buy for your routers.
2. Serviceability check: the carrier verifies both buildings are on-net, near-net, or off-net on their fiber, and engineers a route. For protected service, this is where route diversity gets confirmed — ask for the actual path documentation.
3. Quote and contract: monthly recurring, non-recurring construction or equipment charges, cross-connect responsibilities, term, and the SLA (availability, repair time, and — if offered — latency).
4. Design and provisioning: the carrier assigns optical capacity, builds any needed lateral fiber into the building, and installs transport equipment at each end.
5. Cross-connect ordering: in carrier-neutral facilities, you order the cross-connect from the facility operator separately — this is on you or your advisor, not the carrier.
6. Testing and turn-up: the carrier runs optical tests and presents a clean circuit; you light your side, verify throughput and error rates, and test failover behavior if the service is protected.
7. Acceptance: don't sign acceptance until your own equipment has passed traffic across the link at line rate. Demarc disputes are much easier before acceptance than after.

An advisor earns their keep at steps two through four: getting multiple carriers to engineer the same route, forcing the diversity question in writing, and making sure the cross-connect and construction costs land in the comparison instead of on your first invoice.

Timelines hinge almost entirely on whether fiber already reaches both endpoints. When both buildings are lit and capacity exists on the route, typical intervals run roughly 30–90 days from order to turn-up, varying by carrier and market — optical provisioning is faster than new construction but still involves real engineering, not a self-service checkout. On-net-to-on-net metro circuits at the fast end of that range are common; long-haul and multi-segment routes trend longer.

If either building is near-net, add time for a lateral build — often 60–120 additional days depending on permits, conduit, and whether the route crosses anyone else's property. Off-net builds are project work: three months to a year is a realistic planning range, and some carriers will simply decline or price the construction to make you go away. Protected circuits on routes where a genuinely diverse path requires new construction can double the build scope.

The scheduling trap is the cross-connect: facilities quote their own lead times for physical cross-connects, and a circuit can sit ready for weeks waiting on one. Order cross-connects the day the circuit contract is signed, not after testing.

• Assuming 'protected' means diverse: two paths in the same conduit fail together. Get route maps and compare them; if the carrier won't document diversity, treat the protection as marketing.
• Buying a single unprotected wave for an application that can't go down, then discovering the SLA credits you a few dollars of service instead of your outage losses. SLAs compensate; they don't prevent.
• Wrong optics or interface: ordering a 100G wave and discovering your router needs a different transceiver type, reach, or connector — an expensive, delay-inducing surprise. Match the handoff spec to your hardware before signing.
• Forgetting cross-connect fees: the circuit quote looks great until the facility's monthly cross-connect charge and one-time install fee show up.
• No headroom plan: buying 10G when the replication schedule doubles every 18 months. Mid-term upgrades re-open pricing and term — negotiate the upgrade path (and its price) at signing.
• Treating latency as 'low' instead of a number: if the application needs under 3 ms round trip, specify that and make the carrier commit to a route that delivers it — amplifier huts and route miles add up.
• Skipping the acceptance test: signing off before your own gear has pushed line-rate traffic across the link.
• Designing for one wave with no failover logic on your side: a wavelength is a component — your routers still need to know what to do when it goes down.

1. Are both of my buildings on-net, near-net, or off-net on your fiber — and what does any construction cost?
2. For protected service, are the working and protect paths physically diverse end to end? Will you provide route documentation?
3. What is the guaranteed latency for this route, and is it in the SLA or just a marketing figure?
4. What handoff interface and optics do you deliver, and is the transport equipment included or billed?
5. What's in the SLA: availability percentage, mean time to repair, chronic-outage remedies, and how are credits calculated?
6. What does an upgrade to the next capacity tier cost mid-term, and does it extend the contract?
7. Are there regeneration or amplification points on the route that add latency, and how many route miles is the path?
8. Who is my escalation contact at 2 a.m., and what's the actual NOC process for optical-layer faults?
9. What happens to pricing at renewal — is there an auto-renewal with a rate escalator?
10. If I later want to run multiple protocols or encrypt at Layer 2, is the service fully transparent (jumbo frames, MACsec, my own VLANs)?

The realistic alternatives depend on what problem you're solving. For raw capacity between two points, the comparison is dark fiber, carrier Ethernet, and high-capacity DIA. For 'I need my sites connected and my traffic private,' SD-WAN over internet circuits competes too — at a completely different price and capability tier. The trade-offs:

The most common real-world architecture is a hybrid: a wavelength (or a pair) between the data centers doing the heavy lifting, and SD-WAN over ordinary business internet connecting everything else. The wavelength carries the replication and inter-site backbone traffic; the SD-WAN carries the humans.

Dark fiber deserves a special note because it's the classic wavelength alternative: if both endpoints sit on a route where dark fiber is available and you have (or will hire) the optical engineering skill, lighting your own fiber can beat leased wavelengths economically at very high capacity or over long terms. If you don't have that bench strength, a managed wavelength gets you the same physics with the carrier holding the amplifiers.

Healthcare

Hospital systems replicate electronic health records and imaging between primary and secondary data centers continuously, and a single radiology study can be gigabytes. Wavelengths between facilities and the DR site keep recovery objectives tight and keep that traffic off shared networks. For organizations working within HIPAA security programs, private Layer 1 transport with your own encryption may support the transmission-security controls of a broader compliance program — though no circuit, by itself, makes anything compliant.

Financial services

Trading and payments environments buy wavelengths for the two things they can't get from routed networks: the shortest physical route between two facilities, and latency that doesn't move. Deterministic delay matters as much as low delay — risk models, replication, and order flow all behave better when the network is a constant.

Manufacturing and logistics

Plants increasingly centralize ERP, MES, and quality systems in one data center or colocation facility, which makes the link between plant and data center production-critical. A wavelength pair — or a wave plus a diverse backup — turns 'the network is down, lines are stopped' from an internet-outage gamble into an engineered risk.

WISPs, MSPs, and network operators

Regional operators are wholesale wavelength buyers by nature: 10G and 100G waves to feed tower aggregation sites, connect POPs, and backhaul customer traffic to carrier hotels. For a WISP, a wavelength to a carrier-neutral facility is often the moment the network stops being a collection of radios and starts being a real ISP.

Wavelengths are one of the least commodity products in connectivity: availability depends on whose fiber passes both of your buildings, pricing is engineered per route, and the difference between a good design and an expensive mistake is in the details — diversity, optics, cross-connects, upgrade paths. That's exactly the situation a technology advisor exists for.

TechSellers International is an advisor, not a carrier. We check availability across the providers whose fiber serves both of your endpoints, get multiple carriers to engineer and price the same route, and put the offers side by side — monthly cost, construction, cross-connects, protection, and SLA terms in one comparison. We force the awkward questions (documented route diversity, latency commitments in writing, mid-term upgrade pricing) before you sign rather than after the first outage.

Then we manage the order through installation: construction milestones, cross-connect ordering at the facilities, turn-up scheduling, and acceptance testing with your equipment. And because we're compensated by the providers, the advice and project management don't add a line to your bill — you pay the carrier's price for the circuit, and nothing for the comparison that got you to the right one.