# Point to Point Fiber Network vs PON: Fiber Counts, Equipment, and Fit **Title tag:** Point to Point Fiber Network vs PON: 2026 Comparison **Meta description:** Point to point fiber network vs PON: compare active Ethernet architecture, fiber counts, powered equipment, capacity, and cost drivers for your access build. **Author:** Omar Molina **Published:** November 12, 2026 **Last updated:** November 12, 2026 **Category:** FTTH Design **URL:** https://draftech.com/blog/point-to-point-fiber-network **Primary keyword:** point to point fiber network **Word count:** 3043 **Read time:** 12 minutes ![In an equipment room, a technician plugs a yellow fiber cord into a rack port beside patch panels and a tester.](../../blog/img_point_to_point_fiber_network.webp) --- A dedicated fiber sounds like an easy answer to an operator's capacity problem. At the equipment room, the next question appears: what happens when every customer on that switch sends traffic upstream at once? The access drawing alone cannot answer that. We compare active Ethernet and passive optical networks by tracing the complete service path. Fiber availability matters, but so does the location where powered equipment can be maintained. A spare strand has little value if it ends at a cabinet without a usable port. For an operator choosing an architecture, we want the service requirements on the desk before the cable schedule. That order keeps the design tied to what customers will buy. ## How a point to point fiber network compares with PON A point to point fiber network gives each subscriber a dedicated optical connection to an Ethernet access port. With single-fiber bidirectional optics, that connection uses 1 strand. PON instead connects multiple subscribers through passive splitters to a shared optical line terminal port, changing the feeder allocation and the equipment needed to deliver service. ### Active Ethernet: a separate access port Active Ethernet carries a customer's traffic between powered endpoints over an individual optical link. Calix's Active Ethernet technology overview describes this dedicated subscriber connection and locates access equipment in either a central office or a remote cabinet. The switch location determines how far the individual fibers must travel before traffic joins an aggregation link. Trace the route first. A home-run design brings each connection back to the central site, whereas a remote switch collects the local connections closer to customers and sends their traffic onward over shared uplinks. Both arrangements can deliver point-to-point access. Their feeder requirements differ substantially. We distinguish physical dedication from a service commitment. An exclusive access port can still feed an oversubscribed uplink, and the provider must engineer its upstream network to support whatever bandwidth the contract promises under the agreed traffic conditions. The port label settles only part of that question. ### PON: a shared optical tree In a passive optical network, the optical line terminal, or OLT, communicates with optical network terminals, or ONTs, through an optical distribution network. A splitter divides the optical signal among branches without needing local electrical power. The ONT still needs power. So does the OLT. We place the splitter on the route drawing before comparing strand demand. A splitter close to customers consolidates their traffic onto a shared feeder fiber; a splitter inside the central office leaves individual downstream fibers running much farther toward that office. Calling both designs PON does not make their cable schedules equivalent. The optical tree also establishes a shared capacity domain. Upstream transmissions are coordinated by the PON system rather than sent simultaneously without control. For the placement decisions behind that tree, our [GPON network design guide](/blog/gpon-network-design-guide) explains how splitter arrangement affects the outside plant. Keep the actual branch layout beside the equipment proposal during review. ## Fiber counts follow the switch and splitter locations ### Single-fiber and paired-fiber Ethernet Specify the optics before counting strands. Cisco's SFP Modules for Gigabit Ethernet Applications Data Sheet, document C78-366584, updated October 9, 2026, identifies 1000BASE-BX10-D and 1000BASE-BX10-U as complementary devices operating over a single strand of single-mode fiber. Their transmit and receive wavelengths differ, allowing traffic in both directions on that strand. A paired-fiber Ethernet interface instead uses separate transmit and receive fibers. We record the interface type with the strand assignment so that purchasing cannot quietly replace a bidirectional design with equipment requiring a fiber pair. Connector fit alone proves nothing about optical compatibility. Cisco specifies BX10 operation up to 10 km under the stated conditions. That figure belongs to those optics. It does not establish a universal distance limit for active Ethernet, and we still calculate the installed path's loss against the selected transceivers before accepting a route. ### Count working fibers, then reserve capacity For home-run active Ethernet, every served endpoint needs its own uninterrupted optical path back to an access port. For PON with a field splitter, the feeder serves the splitter and the downstream branches serve individual endpoints. Shared feeder capacity is the source of the strand saving, rather than a disappearance of customer drops. We separate working assignments from growth reserves. Available fibers must reach the place where future demand will appear, and spare capacity behind an inaccessible splice does not provide the same expansion option as a documented path to a usable terminal. This is why [fiber strand count planning](/blog/fiber-strand-count-design-planning) starts with the service geography and the network layers. The comparison below assumes a central access site and field splitting for PON. Moving the Ethernet switch into the field changes both the feeder calculation and the operating responsibility. Recalculate when equipment locations move. | Design question | Home-run active Ethernet | PON with field splitting | | --- | --- | --- | | Subscriber connection | Dedicated optical link to an access port | Branch of a shared optical tree | | Feeder allocation | Individual paths return to the access site | Feeder shared upstream of the splitter | | Central equipment | Subscriber-facing Ethernet ports | OLT ports serving groups of ONTs | | Field power | None between endpoints in this layout | None at passive splitters | | Capacity planning | Access rate and aggregation uplinks | PON tree capacity and aggregation uplinks | | Expansion constraint | Available fibers and access ports | Splitter capacity and optical budget | Cable capacity is only part of the space question. We check whether splice trays and distribution panels can accommodate the proposed assignments without making routine changes difficult. A larger cable entering an enclosure may demand a different closure arrangement even when the conduit has room for it. > Before ordering cable, trace a future customer from the proposed terminal back to the serving equipment. Confirm that the reserved fiber can be connected through every intervening closure without consuming an assignment already promised elsewhere. ## Electronics and bandwidth need separate reviews ### Ports and shared uplinks An Ethernet access switch requires a subscriber-facing optical interface for each dedicated connection. Equipment selection must also account for the switch fabric and uplinks, because adding access ports without increasing upstream capacity can move congestion into the equipment room. We ask for the traffic assumptions behind the uplink choice. PON combines subscriber connections on a shared OLT port through the splitter tree. That reduces central port demand per connected endpoint, but the associated ONTs must work with the selected OLT and its supported service configuration. We review interoperability before treating an available ONT as an interchangeable spare. Count licenses in the equipment schedule. A physically installed port may still require activation, and the commercial terms can change the timing of expansion spending. We ask the supplier to distinguish installed capacity from capacity available for customer service, including any software features required by the operator. ### GPON and XGS-PON capacity ITU-T G.984.2, Gigabit-capable Passive Optical Networks: Physical Media Dependent layer specification, August 2019 edition, covers the familiar GPON rate combination of approximately 2.488 Gbit/s downstream and 1.244 Gbit/s upstream. These are optical line rates. They describe shared transport capacity rather than the throughput promised independently to every attached customer. XGS-PON provides nominal symmetric 10 Gbit/s operation under ITU-T G.9807.1, 10-Gigabit-capable symmetric passive optical network, February 2023 edition, with Amendment 1 dated May 2025. ITU lists both components as in force. Symmetry changes the upstream planning discussion, but the capacity remains shared within the PON tree. We compare committed service demand separately from advertised peak speeds. A residential product allowing bursts and an enterprise circuit requiring sustained throughput impose different planning obligations, even if their sales sheets display the same headline rate. The traffic model must describe that difference. Watch the uplink during testing. An access port can meet its own specification while the path beyond it becomes the bottleneck. Our acceptance plan should therefore test the service at its agreed handoff and load conditions, with the customer equipment's interface included in the review. ### Power and fault exposure A home-run Ethernet layout can keep its access switches in a controlled central facility. A remote-switch design creates another powered site that the operator must reach during an outage. Battery space and heat removal belong in that decision before the cabinet location becomes fixed on the construction drawings. PON's passive distribution plant avoids powering the splitter. It does not eliminate shared failures: damage to a feeder or loss of an OLT port can affect the customers served by that tree. Ethernet also has shared dependencies in its switch and transport path. We identify those dependencies before recommending protection. Our [fiber network design work](/services/fiber-network-design) connects the optical layout to these equipment choices so that a convenient cabinet location does not become an unsupported operating site. The design review needs input from whoever will maintain the network after construction, especially when access depends on a property owner or seasonal conditions. ## Compare costs across construction and operations ### Where the initial budget moves Active Ethernet can increase the number of feeder strands and central terminations in a home-run layout. The resulting cost depends on the route and available infrastructure: adding fibers inside a cable selected for a new build is a different decision from installing another cable through an already crowded duct. Civil work can dominate either option. We therefore compare both architectures against the same surveyed route and construction assumptions, rather than crediting PON with a shorter route or charging Ethernet for a cabinet that its actual design does not require. Otherwise the estimate measures different projects. PON introduces splitter hardware and associated connections while reducing shared feeder demand. Splitters also consume optical power budget, so their arrangement may change the usable reach or optics selection. The component price alone cannot tell an operator whether the proposed split is appropriate for the longest branch. Ask what is already installed. An operator with usable home-run fibers and available switch ports faces a different expansion cost from a builder starting with empty conduit. We would preserve working plant where it supports the required service, then price the specific changes needed to meet future demand. ### What the operator keeps paying for Powered field sites bring continuing obligations. Electricity and backup batteries need an operating budget, and environmental equipment must remain serviceable when a technician is unavailable nearby. A remote switch can be a sound design choice, but its maintenance responsibility has to belong to someone before construction begins. We also consider how new connections are activated. A centrally patched layout may let staff make changes from an equipment room, while a different splitter arrangement can require field access for work that would otherwise be indoors. Neither topology name tells us the full dispatch burden. > Ask the operations lead to walk through an equipment replacement using the proposed drawings. If the replacement requires entry to a locked building, resolve the access arrangement before treating that location as maintainable. A limitation of our preference for dedicated Ethernet on demanding business connections is the extra termination space and port inventory it can require as the customer base grows. We accept that burden when individual connection control earns its place in the service model. For ordinary shared broadband demand, the same burden may add little customer value. Budget the migration too. The first construction package should show how a customer moves to a different access technology without losing the only assigned fiber needed to keep the existing service operating during the cutover. That question can change spare-fiber policy before the initial build. ## Design the upgrade and restoration paths ### Optical margin and acceptance Start with the actual loss budget. We account for the installed fiber length and every optical connection, then include the splitter loss where applicable and the margin the operator requires for future repairs. Equipment reach labels cannot substitute for that calculation. Bidirectional Ethernet needs a compatible optical pair at opposite ends. PON needs a compatible OLT and ONT combination with an optical class suitable for the complete distribution path. In either case, excessive received power can also be a problem; a short link deserves review as much as a long one. The acceptance record should identify the tested endpoints so that measurements can be compared with the assigned path later. We want installation loss results tied to the circuit and supporting fault-location records where the testing specification requires them. An unlabeled measurement is difficult to use during restoration. We keep the equipment configuration beside the physical record. A service failure can result from a provisioning error even when the optical path passes its tests, and replacing a transceiver will not fix a customer mapped to the wrong service profile. Troubleshooting should be able to distinguish those cases without reconstructing the design. ### Expansion without an avoidable rebuild A dedicated fiber can support a later change of endpoint electronics if the fiber path and new equipment meet the revised requirements. We still check port support and optical compatibility before describing that change as straightforward. A retained fiber does not guarantee an unchanged customer device. PON upgrades depend on the chosen migration method. Coexistence equipment can let compatible generations share an optical distribution network, but the additional components and supported wavelengths must fit the loss budget and installed plant. We require a vendor-supported migration design before assuming customers can move without disruption. For a mixed network, reserve dedicated paths where a business service may eventually need them. The same cable can carry fibers serving different architectures. Keep those assignments explicit so that adding a PON branch does not consume the path intended for an enterprise connection. Restoration planning deserves equal attention. Separate fibers in the same cable remain exposed to the same cable cut, and diverse electronics do little for a service whose routes meet at the same vulnerable crossing. We mark common physical dependencies on the route plan before calling a service protected. Document the handoff clearly. The operator should know where responsibility passes from the access provider to the customer's network, including which equipment is included in the service and who can replace it. That boundary affects fault response more directly than the architecture's marketing name. ## Choose a point to point fiber network or PON **Business service operators:** Choose dedicated Ethernet access when individual port control and the required service commitment justify the fiber assignment. Engineer the upstream capacity to match. We would specify protection separately for customers whose outage tolerance requires it, because a dedicated connection by itself supplies no alternate route. **Residential access builders:** Use PON as the starting architecture where shared feeder capacity suits the demand model and passive field distribution reduces operating obligations. Select the PON generation against the planned service tiers and upstream demand. We would revisit the split before promising a service the shared port cannot sustain. **Operators expanding existing plant:** Start with an inventory of usable fibers and powered sites. Retain Ethernet where its existing assignments support the business case, and use PON where consolidation solves a real feeder constraint. A mixed architecture is practical when the records make each customer's serving path unambiguous. Draftech International delivers turnkey telecom and power infrastructure, with engineering 100% in-house and full turnkey construction delivered by Draftech-managed subcontract crews. We carry field survey and planning through construction drawings, permits and agency coordination, then construction, as-builts and closeout. Our [partner and company information](/about) identifies the people responsible for that delivery model; the [project contact page](/contact) provides a place to start a route-specific discussion. If feeder capacity or an uncertain equipment location is holding up your build, we can bring those decisions into a coordinated design package before cable and cabinet commitments become expensive to change. Send the route and proposed service requirements to [info@draftech.com](mailto:info@draftech.com) so we can discuss what needs to be resolved. For a qualifying route, Draftech's [free design offer](/free-design) covers engineering the first 20,000 linear feet at no cost, from feasibility and field survey through permit approval. The owner reviews each request before Draftech commits the package. That review gives us a chance to establish the route's needs and the access architecture the package must support. ## Frequently Asked Questions ### Does active Ethernet always need 2 fiber strands? No. Conventional paired-fiber interfaces use separate transmit and receive strands, but compatible bidirectional optics carry both directions over 1 strand using different wavelengths. Cisco's BX10 data sheet describes this arrangement for its complementary D and U devices. We confirm the selected interface before assigning fibers, because a change in optics can change the strand requirement and the corresponding patching plan. ### Does a dedicated fiber guarantee bandwidth? A dedicated optical link provides an individual access connection, but an advertised 1 Gbit/s port does not establish an end-to-end bandwidth guarantee. The switch uplink and the provider's upstream network still carry shared traffic. We check the service commitment against that full path and define acceptance conditions, rather than treating the access-port rate as proof of sustained throughput to every destination. ### Is GPON the same as XGS-PON? No. GPON commonly uses approximately 2.488 Gbit/s downstream and 1.244 Gbit/s upstream, while XGS-PON provides nominal symmetric 10 Gbit/s operation under their respective ITU-T recommendations cited above. Both share access capacity across an optical tree. We compare their capacity against the proposed service mix and check migration compatibility before assuming an installed GPON endpoint can operate on an XGS-PON service. ### Can Ethernet and PON share the same cable? Yes. Separate fibers within 1 cable can support dedicated Ethernet connections and PON feeder or distribution paths, provided each assignment has compatible equipment and an adequate optical budget. We identify the architecture at the fiber level in the records. Sharing the cable also creates a common physical failure exposure, so services requiring route diversity need a separate assessment of their actual paths. ### Does passive fiber keep working during a power outage? The splitter itself requires no power, but service still depends on powered equipment at both ends: the OLT and the customer's ONT. Losing power at either end interrupts service unless suitable backup is available. For 1 affected customer, we check premises power before assuming a fiber fault; for a wider outage, the serving equipment and shared dependencies need investigation too. ## Related Resources - [FTTH Network Planning](/blog/ftth-network-planning) - FTTH Design - [How to Size Fiber Distribution Hubs for FTTH PON Networks](/blog/fiber-distribution-hub-sizing-ftth-pon) - FTTH Design - [FTTH Low-Level Design](/blog/ftth-lld-splice-point-placement-guide) - FTTH Design - [5 FTTH High-Level Design Mistakes That Cost Fiber Providers Millions](/blog/ftth-hld-design-mistakes) - FTTH Design - [HLD Design Services for FTTH Networks](/blog/hld-design-services-ftth-networks) - FTTH Design - [GPON Network Design: Architecture, Split Ratios, and OSP Planning That Holds Up in the Field](/blog/gpon-network-design-guide) - FTTH DESIGN --- **About Omar Molina:** Leads Draftech's FTTH design production teams, overseeing design quality, deliverable standards, and production capacity. Brings 10+ years leading design production, with expertise in FTTx/GPON architecture, fiber construction packages, and AutoCAD production workflows. [info@draftech.com](mailto:info@draftech.com)