Transmission Media & Legacy Transport
Last updated 12 Oct 2026
At a glance
Most large estates run fibre, microwave and satellite together, plus a long tail of SDH, SONET and older optical. Each medium reaches the orchestrator through a different integration path. The catalog should treat the medium as a resource attribute and SLA constraint, not a separate product, and plan for legacy transport that has no service abstraction at all.
The picture
One service, several media. The medium changes the integration path, not the product.
1 · One service
CFS:Internet-Access
100 Mbps · remote site
SLA constraints: max latency, availability
Medium is a resource attribute. The SLA rules out media that cannot meet it.
2 · Medium and its integration path
- Fibre / DWDM
Optical controller · NETCONF/YANG, OpenROADM
Mature. Stable service abstractions.
- Microwave
Vendor controller · NETCONF (modern) or proprietary EMS
Rain fade and line of sight leak into capacity and assurance.
- Satellite, operator-owned
Vendor NMS · SOAP or proprietary REST
Bandwidth on demand, beam handover, weather-aware policy.
- Satellite, third-party operator
Provider API · often through automation middleware
Proprietary, frequently versioned, with waiting states.
3 · Legacy transport
SDH / SONET, pre-OpenROADM optical
EMS exposes element-level operations only
- create cross-connect
- configure port
Adapter
Invents the service model and translates it over SOAP, CORBA or a proprietary NBI. Rework on every EMS upgrade.
How it works
The three media
Geography, regulation, time to market and cost decide the mix more than preference. The comparison below is about the properties that change orchestration and SLA design.
| Property | Satellite | Fibre | Microwave |
|---|---|---|---|
| Throughput | 10 Mbps – 300 Mbps (up to ~1 Gbps modern LEO) | 1 Gbps – 100+ Gbps (Tbps with DWDM) | 100 Mbps – 10 Gbps |
| Latency | GEO 500 – 700 ms; LEO 20 – 50 ms | ~5 – 10 ms per 1000 km | A few ms, comparable to fibre |
| Reliability | Moderate: rain fade, atmospheric conditions | Very high; weather-independent | Moderate: rain at higher frequencies; needs line of sight |
| Coverage | Global; suits remote areas | Limited to the cable footprint | Typically 30 – 50 km per hop |
| Deployment speed | Moderate | Slow: civil works | Fast |
| Cost shape | High equipment and operating cost | High capex, low opex | Moderate |
| Typical use | Remote, maritime, aviation, disaster recovery | Backbones, core networks, data centres | Cellular backhaul, enterprise links, temporary links |
Integration shape per medium
| Medium | Managed through | Protocol | Orchestration challenge |
|---|---|---|---|
| Fibre / DWDM (modern) | Optical domain controllers, OpenROADM-aligned | NETCONF/YANG | Mature; legacy SDH is the residual gap |
| Microwave (modern) | Vendor microwave controllers | NETCONF/YANG, or proprietary EMS for older kit | Per-hop link budget (rain fade, line of sight) leaks into capacity planning and assurance |
| Satellite, operator-owned | Vendor satellite NMS | SOAP or proprietary REST; rarely NETCONF | Bandwidth on demand, beam handover, weather-aware policy; session-style integration |
| Satellite, third-party operator | Provider-owned; operator integrates by API | Vendor REST with waiting states | Proprietary, frequently versioned APIs; often wrapped in automation middleware |
Modelling the medium
A "100 Mbps Internet at a remote site" service may be deliverable over fibre, microwave or satellite, with different latency, availability and cost. Model the medium as a resource property and an SLA constraint. Three products for three media force the customer to choose something they should not need to know about, and force a catalog refactor every time a new medium becomes viable, as LEO satellite did from 2022 onward.
Legacy transport
Large estates still carry substantial SDH, SONET and pre-NETCONF OTN. These elements predate model-driven management, and their EMS were not built to expose service abstractions. They are the largest single source of "we cannot fully orchestrate this domain" caveats in transformation programmes.
- The abstraction gap. The EMS often exposes only element-level operations: create cross-connect, configure port. The layer above must invent its own service model and translate it over SOAP, CORBA or a proprietary interface. That is months of adapter work per platform, repeated on each EMS upgrade.
- Generations matter. Modern optical platforms (post-2015 ROADM, OpenROADM-aligned) speak NETCONF/YANG. Legacy SDH/SONET and pre-OpenROADM optical do not. A plan that assumes "optical means NETCONF" because the new kit does will meet the legacy footprint inherited from acquisitions and long-life backbone.
Who owns what
| Entity | System of Record | System of Engagement | System of Reference | Notes |
|---|---|---|---|---|
| Transmission medium of a delivered service | Resource Inventory | — | Service Inventory, assurance | A resource attribute, set at design and assign; assurance needs it to interpret weather-related degradation. |
| Medium-dependent SLA constraints (latency, availability) | Service Catalog | — | SOM | Rules out media that cannot meet the SLA; does not name the medium as a product. |
| Third-party satellite capacity | Satellite network operator | Operator automation / ROM | Resource Inventory | Provider-owned; the operator holds a reference and the order state. |
| Legacy SDH/SONET circuits | Legacy EMS (often in practice) | — | Resource Inventory | Inventory should be the SoR; in many estates the EMS is the only reliable record. |
In practice
Most large networks combine all three media: fibre for capacity, microwave for fast deployment and backhaul, satellite for reach and resilience. Optical domain controllers cover modern DWDM well. Microwave and operator-owned satellite still lean on vendor NMS. Third-party satellite capacity is usually integrated through automation middleware, because the provider APIs are proprietary and change often.
The trade-off
What it solves
Treating the medium as a resource attribute lets one product be delivered over whichever medium meets its SLA, and lets new media be added without new products.
What it does not solve
- The physics. Satellite latency and microwave rain fade still constrain what an SLA can promise.
- Legacy transport with no service abstraction; that still needs adapters.
- Proprietary third-party satellite APIs that change on the provider’s schedule.
What it adds
- SOM needs medium-selection rules tied to SLA constraints and serviceability.
- Assurance must interpret medium-specific degradation, such as weather on microwave and satellite.
- Adapters for legacy EMS have to be maintained through every EMS upgrade.
Architecture decision
What should happen to legacy SDH/SONET services in the target architecture?
Model and orchestrate them through adapters
Choose when the circuits carry long-term regulated or contracted services and the volume justifies months of adapter work.
Represent them in the catalog, fulfil manually
Choose when change volume is low, the services must stay visible commercially, and automation would not pay back.
Migrate the services and retire the platform
Choose when contracts allow it, there is a dated plan with funding, and customers can be moved to modern equivalents.
Breaks if retirement is assumed without a funded plan. The services outlive the assumption, have no clean model, and run on workarounds and a parallel commercial record.
Takeaways
Key Takeaways
- Fibre gives capacity, microwave gives speed of deployment, satellite gives reach; most large estates run all three.
- Each medium reaches the orchestrator through a different controller, protocol and set of waiting states.
- Model the medium as a resource attribute and an SLA constraint; one product covers every medium.
- Legacy SDH/SONET and pre-OpenROADM optical have no service abstraction; adapters must invent one.
- Plan for legacy transport as long-lived unless a funded retirement plan exists.
Check yourself
Sales asks for three products: "Remote Internet – Fibre", "– Microwave" and "– Satellite". What do you propose instead?Show answerHide
One product with SLA attributes (latency, availability). The medium is chosen at design and assign as a resource attribute, constrained by the SLA. A new medium then needs no new product.
Why can an orchestrator not simply call the EMS of a legacy SDH platform to create a circuit?Show answerHide
The EMS exposes only element-level operations (cross-connects, ports). Something above it has to own a service model and translate it into element steps, and keep that adapter working through EMS upgrades.