Microsoft Teams

Calls disconnect at almost exactly 15 minutes when routed through an on-prem SBC.

Learn how to design and deploy **Microsoft Teams Direct Routing with Local Media Optimization (LMO)** to keep RTP audio streams within your local corporate network. This guide covers the signaling and media path separation, central versus downstream SBC topologies, PowerShell configuration, and SDP offer/answer exchanges. Master LMO architecture to reduce internet bandwidth usage, lower latency, and maintain strict compliance across multi-site enterprise environments.

## Architecture Overview In Microsoft Teams Direct Routing with **Media Bypass** enabled, Teams clients attempt to establish media paths directly with the **Session Border Controller (SBC)** using Interactive Connectivity Establishment (**ICE**). This bypasses the Microsoft Cloud Media Proxies to minimize latency and optimize audio quality. ## ICE Negotiation Failure When the **ICE candidate exchange** fails during SIP signaling, the Teams client and the SBC cannot agree on a mutually reachable UDP transport path. Consequently, the call connects signaling-wise, but media setup times out, resulting in immediate **one-way/no-way audio** followed by an automatic call termination after **10 seconds**.

## TLS Handshake Failure Overview Inbound and outbound voice calls between the on-premises **Session Border Controller (SBC)** and Microsoft Teams Direct Routing fail during the **SIP TLS 1.2/1.3 renegotiation** or initial transport establishment. ## Traffic Flow Impact The SBC receives `TLS alert level: fatal, description: unknown CA` or `bad certificate` from Microsoft SIP proxies (`sip.pstnhub.microsoft.com`), preventing any SIP `INVITE` or `OPTIONS` messages from being processed successfully.