Overview

As a network architect or telecom engineer, granular control over subscriber bandwidth is non-negotiable. While many administrators rely on per-subscriber traffic shaping via RADIUS or BRAS, the most efficient and hardware-native approach is to implement bandwidth limiting directly within the OLT (Optical Line Terminal) service profile. This FAQ addresses the most critical pre-sales and post-sales technical questions regarding bandwidth management per user directly on the OLT, from CLI configuration to optical module compatibility and hardware failover.

Frequently Asked Questions

Q1: What is the exact CLI command sequence to limit bandwidth per user within an OLT service profile?

To limit bandwidth per user, you must configure traffic profiles and apply them to the GPON ONT service. The definitive CLI sequence is: `configure traffic-profile committed-information-rate peak-information-rate ` to define the rates, followed by `configure service-port user profile ` to bind it to the user's ONT. For example, to set a strict 100 Mbps CIR with a 150 Mbps PIR, use `committed-information-rate 100000` and `peak-information-rate 150000`. This ensures that the hardware enforces the bandwidth limit at the forwarding plane without taxing the control CPU.

Q2: Does limiting bandwidth per user via the OLT profile affect the maximum switching throughput or backplane capacity?

No, bandwidth limiting does not reduce the total maximum switching capacity of the OLT chassis. The backplane capacity (e.g., 3.2 Tbps) and the line-rate forwarding performance (e.g., 960 Mpps) are fixed hardware capabilities. Bandwidth profiles only apply policing and shaping to specific GEM ports, which is a hardware-level function on the PON MAC. This means you can simultaneously enforce thousands of unique user profiles without degrading the OLT's overall non-blocking throughput.

Q3: Are there specific SFP/SFP+ optical module compatibility requirements for the uplink interface when enabling per-user bandwidth shaping?

Bandwidth limiting itself does not alter SFP compatibility requirements, but the uplink module must support the necessary speeds to handle aggregated user traffic. For 1G uplinks, use standard SFP modules (e.g., 1000BASE-SX/LX). For 10G uplinks, use SFP+ modules (e.g., SR/LR). Crucially, if you are shaping hundreds of users to 1 Gbps each, ensure your uplink SFP+ module supports the resulting aggregate throughput. Always use modules that adhere to MSA (Multi-Source Agreement) standards; third-party optics generally work, but we recommend referencing the OLT's approved optical compatibility matrix for optimal signal integrity and DDM monitoring.

Q4: How does the OLT prioritize bandwidth when multiple profiles (e.g., SLA vs. Best-Effort) are applied to users on the same PON port?

The OLT uses a strict hierarchical QoS scheduler. By assigning different priority queues (e.g., using the `priority` parameter within the traffic profile), the system guarantees bandwidth for high-SLA profiles before allocating remaining bandwidth to best-effort profiles. When multiple users share the same PON port, the total bandwidth consumed cannot exceed the physical PON interface speed (e.g., 2.5 Gbps downlink). The OLT dynamically manages this using Dynamic Bandwidth Allocation (DBA), ensuring that high-priority profiles receive their CIR even during congestion, while best-effort users share the leftover PIR bandwidth fairly.

Q5: What happens to per-user bandwidth limits during a redundant power supply failover event?

In the event of a primary AC/DC power supply failure, the OLT will perform a hitless failover to the redundant power module. The bandwidth limitations configured in the subscriber profiles are stored in non-volatile memory and applied by the hardware ASICs. This means that during and after the failover, the bandwidth policies are preserved and enforced immediately, requiring no configuration synchronization. The failover time is typically under 50ms (telecom-grade), which is insufficient to disrupt existing traffic flows or the rate-limiting hardware functions.

Q6: Is there a difference in the bandwidth-limiting mechanism between the current generation OLT and previous-generation models?

Yes, a primary difference lies in the granularity and depth of the traffic queues. Previous-generation OLTs often supported limited profile scaling and shallow buffer depths, which could lead to tail drops during micro-bursts. Current-generation OLTs feature intelligent hierarchical scheduling with a significantly larger packet buffer (e.g., 36 MB shared buffer vs. 4 MB previously) and support for 8 hardware queues per port. This allows for much more precise shaping and policing at the user level, preventing head-of-line blocking and ensuring fair access to the upstream bandwidth of the PON port.

Q7: Can I dynamically change a user's bandwidth profile without rebooting the ONT or dropping their PPPoE session?

Yes, you can dynamically modify the bandwidth profile on the fly by using the command `configure traffic-profile committed-information-rate ` followed by `commit`. The OLT instantly re-syncs the new parameters with the ONT via the OMCI (ONT Management and Control Interface). This change does not require an ONT reboot and will not drop the user's PPPoE or DHCP sessions. The new rate limit takes effect within the next DBA cycle, usually within 2 to 5 seconds, ensuring a seamless adjustment to the subscriber's service level.

Q8: How can I verify that the bandwidth limit per user is effectively applied after CLI configuration?

You can verify the bandwidth policy using the `show traffic-profile ` command to review the CIR and PIR parameters. For operational verification, use `show service-port ` to confirm the profile binding. For real-time traffic monitoring, utilize `show port statistic` on the specific PON interface or perform an iPerf test from the subscriber side while monitoring the OLT's traffic statistics to confirm the limiting is taking place at the configured thresholds. Additionally, you can use `show gpon statistics` to view drop counts specific to the policed traffic.