Lists the key hardware and scalability limitations of the VOQ model.
Before you install the QoS features on your router, consider these limitations that could impact scalability, system memory usage, and system behavior under high-load conditions:
-
Replication overhead: Each egress queue must be replicated as an ingress VOQ on every slice of every NPU or ASIC. This increases memory usage significantly.
-
Reduced scalability: The total egress queue scale is lower due to the replication requirement, limiting system-wide queueing flexibility.
-
Resource consumption: As the number of egress ports or linecards increases, the system must replicate VOQs across all ingress NPUs, consuming significant buffer and memory resources.
For example, adding a new NPU with 20 interfaces may require 160 additional VOQs on every existing NPU (20 interfaces × 8 traffic classes per interface).
-
On-chip buffer usage: In high-scale scenarios with 1000+ active VOQs, on-chip buffer (OCB) may be exhausted, causing unexpected traffic drops even when traffic rates are within shaping thresholds.
-
CGM profile scaling during policy modification: The system supports 29 congestion management (CGM) profiles per router. During in-place modification of queuing policies—such as changing shaper parameters in parent policies—the system may temporarily require additional CGM profiles. If the required profiles exceed the supported limit, the configuration commit fails due to resource exhaustion.
Plan CGM profile usage per NPU to prevent exceeding hardware resource limits.