UC LabFree University

T5 · T5-R02

LLQ and the delay budget

Learn it in 5 minutes

  • Priority queue bandwidth is a limiter for the EF class.
  • Serialization, queuing, propagation, packetization, and jitter buffer all add delay.
  • ITU-T G.114 is the usual citation for a 150 ms one-way planning target.
  • A queue cannot create bandwidth the circuit does not have.
  • Confirm the current ITU text before you treat 150 ms as a contract.

Visual

LLQ and the delay budget diagram

What is going on

G.711 at 20 ms is about 80 kbit/s at layer 3 and about 87.2 kbit/s with an 18-byte Ethernet header. The formula is (payload + 40 + layer-2) times 8 times packets per second, then divide by 1000 for kbit/s. Size the priority class from calls you allow, not from a feeling.

Confirm before relying. Confirm the current G.114 text. The 150 ms figure is the classic planning number, not a promise about your carrier.

Worked example

Twenty G.711 calls want about 1.6 Mbit/s at layer 3 before overhead. A priority statement of 512 kbit/s will drop or delay the rest. Users hear it as clipping, not as a SIP error.

Quiz

1. LLQ adds bandwidth to a circuit.

Ecosystem

Sister pieces sit on the ecosystem page: the glossary, the mind map, the integration map, and the sibling labs.

  • LLQ — A queue that serves a priority class up to a configured rate.
  • Packetization — The time to fill one RTP payload.

Config. Write the class rate next to the number of calls.

Troubleshooting. If signaling succeeds and MOS collapses only at busy hour, look at drops in the priority class.

Bridge. CAC in CUCM can limit calls so the queue is not overrun. QoS and CAC are partners.

Mind map, sister technologies, and future hooks live on the ecosystem page.

Sources

Product screens move. Confirm the guide for your release before you change a live system.

Sources and go deeper

Opened 2026-10. A row marked confirm has no link, because that page was not verified.

sources.json Markdown