The inter-building fiber backbone is the hardest part of a campus network to change. Switches are renewed every few years; replacing a buried cable means excavation, permits and downtime. In a factory, hospital, campus or logistics center, a wrongly chosen fiber count or two “redundant” cables running through the same trench remain a risk carried for years. Below we address topology, route diversity, fiber count calculation, outdoor cable types and building entry in decision order. The basic comparison of OM3, OM4 and OS2 is the subject of a separate article.
Topology options for a campus backbone
Physical cable topology and logical network topology aren’t the same thing; but the first sets the limits of the second. There are three basic options:
| Topology | Structure | Advantage | Limitation |
|---|---|---|---|
| Star | Each building connects to the central main distribution with a single cable | Simple, easy to manage, fiber count clear per building | Single route, single point of failure |
| Dual-route star | Each building connects to the center with two physically separate cables | Resilient to route failure | High excavation and cable cost |
| Physical ring | The cable passes through the buildings in turn and returns to the center | Two-way access to each building, shorter total excavation | Many splice points, complex fiber planning |
In enterprise networks, the backbone’s logical structure is usually a hierarchical star: each building’s distribution switch connects to the core with two uplinks. Even if a physical ring is chosen, the fibers in the ring are spliced in splice enclosures so that each building connects to the center from two different directions. The logical network remains a star, and dependence on loop prevention protocols is reduced. For the architectural choices of the core and distribution layers, see the three-tier and collapsed core campus network article.
For applications such as cameras, barriers and perimeter security, a ring can also be used at the logical level; but that is a separate design problem and shouldn’t be mixed with the enterprise data backbone on the same fibers. When they’re mixed, troubleshooting gets harder.
Physical route diversity and single points of failure
Running two uplinks over two different fibers of the same cable provides logical redundancy, not physical redundancy. An excavator, a fire in a duct or a rodent cuts both fibers at once. This is a “redundant” design you may encounter: two ports in the configuration, LACP or two separate uplinks; a single cable under the ground. A single cable is a single point of failure. Conditions to look for real redundancy:
- Separate conduit/duct lines: The two cables shouldn’t run in the same trench or the same conduit bundle; if possible, they should enter the building from different sides.
- Separate building entrances: The same entry point and the same splice enclosure reset route diversity to zero at a single point.
- Separate termination points: At the center, the two cables should terminate on different ODFs or at least different panels, and connect to different switches on the active side.
- Documenting common points: Unavoidable common points such as a bridge crossing, a railway underpass or a single entrance gate should be recorded as risks.
Running dual routes to every building isn’t always economical. The building’s business continuity needs make the decision: a building housing a production line or a data center deserves a dual route; an auxiliary building used as a warehouse can manage with a single route and a wireless backup.
Fiber count calculation and spare fiber margin: an example
Calculate the fiber count not as “the standard 12 fibers” but from a needs list per building. The worked calculation below shows the method for a production building; the service list changes in every project.
| Service | Connectivity | Fibers |
|---|---|---|
| Enterprise network uplinks | 2 duplex links (to different core switches) | 4 |
| Camera network | 1 duplex link | 2 |
| Building automation and access control | 1 duplex link | 2 |
| OT / production network | 1 duplex link | 2 |
| Carrier or dedicated line handoff | 1 duplex link | 2 |
| Today’s need | 12 |
The following margins are added on top of this total:
- Fault and repair margin: Spare fibers to switch to immediately when a fiber in the cable is damaged.
- New service margin: Networks to be separated in the future, or a new tenant/unit.
- Parallel optic margin: At 40G or 100G, some interfaces (e.g. the SR4 type) use four transmit and four receive fibers, i.e. 8 fibers per link. On single-mode, most long-reach interfaces run duplex; however, breakout scenarios may require parallel fibers.
In the example, adding an equal amount of margin to today’s 12 fibers gives 24 fibers. This fills a standard 24-fiber cable right to the limit; a deliberate choice between 24 and 48 fibers is made based on the building’s growth expectations, and 48 fibers is usually the safer choice for critical buildings. In a dual-route design, each cable must be able to carry all critical services alone; do the calculation separately for each route. Cable cost is small next to excavation, conduit and labor. Skimping on fibers isn’t a saving.
Outdoor fiber cable types
Outdoor cable is chosen according to conditions such as water, moisture, temperature changes, mechanical load and rodents. The main constructions:
| Cable construction | Feature | Typical use |
|---|---|---|
| Loose tube, gel-filled | Fibers in gel-filled tubes; longitudinal water protection | In conduit, underground duct |
| Loose tube, dry water-blocked | Water-swellable yarn/tape instead of gel; cleaner splice preparation | In conduit, routes with many splice enclosures |
| Steel armored (corrugated tape) | High protection against rodents and crushing; contains metal, requires bonding | Direct burial, areas with high rodent risk |
| Dielectric rodent-protected | Glass yarn or similar non-metallic protection; requires no grounding | Open areas with lightning risk |
| ADSS aerial cable | Self-supporting, all-dielectric | Pole line |
| Indoor/outdoor | Jacket resistant to the outdoor environment and suitable for the indoor fire rating | Transition at the building entrance without a splice |
Metal-armored cables are connected to the telecommunications bonding network at the building entrance; details are in the telecommunications grounding and bonding article. In open sites with high lightning risk, dielectric cable is the easiest way to preserve galvanic isolation between buildings.
Along the route, also check: the protective conduit (e.g. HDPE) and subducts inside it, splice/pull chambers at set intervals and at changes of direction, warning tape above the conduit, the manufacturer’s maximum pulling force and minimum bend radius (different values are given for during pulling and after installation). Pulling force exceeded during installation comes back as loss that isn’t visible that day but grows later. Leaving an empty subduct is the cheapest way to pull a second cable without excavation.
Building entry, splice enclosure and termination
Outdoor cables mostly don’t meet in-building fire rating requirements. In Europe and Turkey, indoor cables carry a reaction-to-fire class under the EU Construction Products Regulation (CPR, 305/2011) and the national regulation aligned with it; the class required in the project is determined by fire regulations and the specification. There are two paths:
- Transition at the entry point: The outdoor cable is terminated in a splice enclosure or entrance panel near the building entrance, and continues inside the building with indoor cable of the appropriate fire class.
- Indoor/outdoor cable: If the cable meets the requirements of both environments, it can be pulled to the server room without a splice; the number of splice points and losses decreases.
For termination, fusion splicing factory-terminated pigtails is the standard method in enterprise projects. Splices are stored neatly in an ODF or a fiber patch panel with splice cassettes; each cassette is labeled with the cable and fiber range it belongs to. For choosing connector type and polish, see the fiber connector selection article; for post-installation acceptance measurements and loss budget calculation, see the fiber optic testing and OTDR article. Ask for both OLTS and OTDR measurements on backbone lines; splice points and possible bends are recorded, and years later when you’re hunting for a fault you have a reference trace. If the campus backbone will also carry remote security cameras, calculate recording traffic and endpoints separately for the fiber optic camera network infrastructure.
Preparing for future speed increases
Backbone cable is chosen to carry several generations of interfaces that will run over it. The reach of IEEE 802.3 interfaces on multimode fiber shrinks noticeably as speed increases: while 10GBASE-SR is defined up to 300 m on OM3 and 400 m on OM4, for 40GBASE-SR4 this distance is 100 m for OM3 and 150 m for OM4; and for 100GBASE-SR4 it’s 70 m for OM3 and 100 m for OM4. On the single-mode side, interfaces such as 10GBASE-LR and 100GBASE-LR4 are defined up to 10 km. The difference is clear. The detailed characteristics of ITU-T G.652 compliant single-mode fiber are in the ITU-T G.652 recommendation.
Practical rules for a new inter-building backbone:
- OS2 single-mode is the basis on new routes; existing multimode lines can be kept if the distance is sufficient.
- In critical buildings, the fiber count is set to meet parallel optic or additional service needs at the next speed migration.
- Empty subduct and chamber capacity are left for a second cable.
- The fiber assignment table, OTDR traces and route drawings are kept as as-built records.
Backbone design decision table
| Question | Answer | Design decision |
|---|---|---|
| Is the building critical for business continuity? | Yes | Dual route, separate entrance, separate core switch |
| Are the buildings lined up along a perimeter? | Yes | Physical ring, logical star fiber assignment |
| Does the distance exceed a few hundred meters? | Yes | OS2 singlemode |
| Is there a risk of rodents or crushing along the route? | Yes | Armored or dielectric rodent-protected cable |
| Is lightning risk high in an open site? | Yes | Dielectric cable, avoiding metal elements |
| Is the in-building distance long? | Yes | Transition to indoor cable at the entrance, or indoor/outdoor cable |
A campus backbone makes sense when designed together with the active network architecture and the physical infrastructure. For architecture and redundancy decisions, see our enterprise network consulting service page; for route, pulling and termination work, our structured cabling and low-voltage infrastructure service page.
Frequently asked questions
Should multimode or single-mode be used for an inter-building backbone?
At campus distances, the default choice is OS2 single-mode fiber. With multimode fiber, reach at 10G and higher speeds is limited to a few hundred meters and shrinks as speed increases; single-mode supports future 25G, 100G and higher interfaces over kilometers with the same cable. If existing multimode infrastructure will be kept, adding single-mode on new routes is the most flexible path.
When does a ring topology make sense on an enterprise campus?
If buildings are lined up along a perimeter road and running two separate routes from the center to each building is expensive, a physical ring provides two-way access. The logical design is still mostly a star: the fibers in the ring are spliced so that each building connects to the center from two different directions.
How much spare fiber should be allowed?
Calculate it not with a fixed ratio but with a list of needs: add up the fibers in use today, planned services and parallel optic needs, then go up to the next standard fiber count considering the cost and time of replacing the cable. Because excavation and pulling labor are much more expensive than the fiber count, a high margin is often economical.
How far can outdoor fiber cable run inside a building?
Outdoor cables usually don’t meet in-building fire rating requirements. That’s why at the building entrance there’s a transition to indoor or indoor/outdoor cable through a splice enclosure or entrance panel. The permitted distance and cable class are determined by local fire regulations and the project specification.