Introduction: The Paradox of Normal Optical Power with Zero Throughput

In the domain of Passive Optical Networks (PON), few issues are as perplexing to field engineers and NOC teams as an ONT reporting optimal optical power levels—typically between -8 dBm and -27 dBm per ITU-T G.984 (GPON) or ITU-T G.9807 (XGS-PON) standards—while user traffic yields 0 Mbps throughput and a complete lack of internet connectivity. This scenario defies the common assumption that acceptable Rx/Tx optical power guarantees a functional link. The root cause almost always lies not in the physical layer, but in Layer 2/3 provisioning, ONT registration states, VLAN mismatches, PPPoE/DHCP failures, or OLT port isolation. This manual dissects the ONT data path, provides a systematic configuration and troubleshooting workflow, and equips enterprise and ISP engineers with the metrics and commands needed to restore line-rate forwarding.

Core Architecture & Switching Logic: Why Optical Power Is Not Enough

ONT Data Path and Packet Pipeline

An ONT is not a simple media converter. It contains a PON MAC, a switch fabric (often with 1–10 Gbps capacity), 802.1Q VLAN engines, QoS schedulers, and a CPU for management. The optical transceiver may report normal Rx power (e.g., -18.5 dBm) and Tx power (e.g., +2.1 dBm), but if the ONT is in O5.1 (Operational) state without OMCI configuration, or if the GEM port is not mapped to a TCONT, the Ethernet side will pass no frames. Common culprits include: missing VLAN, wrong PPPoE credentials, DHCP snooping blocking, MAC address filtering, storm control shutting the port, or ONT in bridge mode without NAT. The optical layer is healthy; the service layer is dead.

Layer 2/3 Configuration Faults

Per IEEE 802.1Q, the ONT must tag or untag frames correctly. A VLAN mismatch between the ONT and OLT (e.g., ONT expects VLAN 100, OLT sends VLAN 200) results in 0 throughput despite normal optical power. Similarly, PPPoE discovery may fail if the ONT is not in route mode or if MTU is misconfigured (e.g., 1492 vs 1500). DHCP Option 82 insertion can also cause IP lease failures. The ONT may show link up on the Ethernet port, but ARP and ping fail. This is a configuration issue, not an optical one.

OLT and Uplink Bottlenecks

On the OLT side, traffic shaping, DBA (Dynamic Bandwidth Allocation) misconfiguration, or ONU TCONT allocation can starve the ONT. If the OLT uplink is congested or STP blocked, throughput drops to zero. Check OLT MAC table and ONT registration status via OMCI.

Technical Specifications: Key Metrics for ONT Throughput Diagnosis

The following table lists the critical parameters to verify when optical power is normal but throughput is zero. Values are typical for GPON/XGS-PON enterprise ONTs.

Configuration & Optimization Manual: Step-by-Step Troubleshooting

Step 1: Verify ONT Registration and OMCI State

Access the ONT via Telnet/SSH/Web. Check OMCI status: it must be O5.1 (Operational). If O5.2 (Disabling) or O7 (Emergency Stop), the ONT is not provisioned. Use OLT commands like show ont info or show gpon ont.

Step 2: Validate VLAN and Service Profile

Confirm VLAN ID, 802.1p priority, and GEM port mapping. Mismatched VLAN is the #1 cause of zero throughput with normal optical power. Use show vlan and show gemport.

Step 3: Check PPPoE/DHCP and IP Layer

If PPPoE, verify username/password and service name. If DHCP, check Option 60/82. Run ipconfig /renew or dhclient -v. MTU must be ≤ 1492 for PPPoE.

Step 4: Buffer Allocation and QoS

For low-latency applications, ensure ONT buffer is not exhausted. QoS shaping on OLT must match ONT TCONT. Check DBA thresholds.

Step 5: Physical and MAC Layer Checks

Even with normal optical power, a dirty fiber or bad splice can cause high BER (e.g., >10^-6), leading to frame drops. Check FEC counters and BIP errors. Replace patch cords and clean SC/APC connectors.

Step 6: OLT Uplink and Ring Topology

In ring or tree topologies, a loop or STP blocking can isolate the ONT. Verify RSTP/MSTP states. For leaf-spine aggregation, check ECMP hashing.

Deployment Scenarios: Enterprise and ISP Case Studies

In a 1000-unit campus deployment, 12 ONTs showed -19 dBm Rx but 0 Mbps. Root cause: OLT VLAN pool exhaustion and PPPoE session limit per MAC. After re-provisioning VLANs and increasing PPPoE max-sessions to 4, throughput reached 940 Mbps (line rate for 1G). In another ISP case, ONT OMCI was O5.1 but GEM port was unmapped; fixing TCONT allocation restored 2.5 Gbps XGS-PON service. Always validate service profile before replacing hardware.

Conclusion

Normal optical power on an ONT is a necessary but insufficient condition for internet throughput. The fault domain shifts to Layer 2/3 provisioning, OMCI state, VLAN, PPPoE/DHCP, QoS, and OLT uplink configuration. By following the structured troubleshooting manual above—verifying OMCI O5.1, VLAN mapping, PPPoE credentials, MTU, DBA, and FEC counters—engineers can restore line-rate forwarding without unnecessary truck rolls. Remember: optical power is only the first mile; throughput depends on the entire packet pipeline.