Architecting Session Timer Configuration to Prevent Orphaned Call Sessions on Trunks

How-To-Articles
1

Architecting Session Timer Configuration to Prevent Orphaned Call Sessions on Trunks

What This Guide Covers

This guide details the precise configuration of SIP Session Timers and TCP Keepalive intervals on Genesys Cloud CX and NICE CXone trunks to prevent state desynchronization between the CCaaS platform and upstream carrier infrastructure. The end result is a resilient telephony edge where silent legs, network drops, and stale media streams are automatically detected and torn down, eliminating orphaned sessions that consume concurrent channel capacity and inflate carrier billing.

Prerequisites, Roles & Licensing

  • Licensing Tier: Genesys Cloud CX Standard or higher (Telephony Edge access required). NICE CXone Standard or higher.
  • Granular Permission Strings:
    • Genesys: Telephony > Trunk > Edit, Telephony > Trunk > View
    • NICE CXone: Admin > Telephony > Trunk Management
  • OAuth Scopes: telephony:trunk:write, telephony:trunk:read
  • External Dependencies: Direct peering with a SIP carrier that supports RFC 4028 (Session Timer Extension) or a SIP trunk provider that exposes advanced SIP header configuration.

The Implementation Deep-Dive

1. Understanding the Desynchronization Failure Mode

Before configuring timers, you must understand why orphaned sessions occur. A SIP session consists of three distinct states: signaling state, media state, and application state.

The signaling state is managed by the SIP stack. The media state is managed by RTP/RTCP. The application state is managed by the CCaaS platform (Genesys Cloud or NICE CXone).

Orphaned sessions occur when these three states diverge. This typically happens in two scenarios:

  1. Silent Network Drops: The underlying IP network drops packets due to congestion, firewall timeout, or NAT expiration. The SIP endpoints (the CCaaS edge and the carrier gateway) stop receiving packets, but neither side receives a formal SIP BYE or CANCEL. Both sides assume the call is active.
  2. Stale Media: The media stream stops flowing due to codec negotiation failure or one-way audio issues, but the signaling channel remains open. The carrier continues to bill for the duration of the signaling connection.

In a high-concurrency environment, orphaned sessions consume valuable concurrent channel licenses. If a carrier provisions 100 channels and 10 are orphaned, your effective capacity is 90. Under load, this leads to forced call drops and degraded service levels.

The Trap: Relying solely on the SIP Expires header or TCP Keepalives is insufficient. TCP Keepalives detect network connectivity but do not verify media health. SIP Expires headers are often ignored by legacy carrier gateways. You must implement a layered defense using Session Timers for signaling validation and RTCP/RTCP-XR for media validation.

2. Configuring SIP Session Timers (RFC 4028)

SIP Session Timers provide a mechanism for endpoints to negotiate a maximum session duration and exchange periodic re-INVITE or UPDATE requests to confirm the session is still active.

Genesys Cloud CX Configuration

In Genesys Cloud, Session Timer support is enabled at the Trunk level.

  1. Navigate to Admin > Telephony > Trunks.
  2. Select the target trunk and click Edit.
  3. Expand the Advanced Settings section.
  4. Locate the Session Timer configuration block.
  5. Set Session Timer Support to Enabled.
  6. Set Session Timer Interval to 180 seconds. This is the standard industry value. Values below 120 seconds increase signaling overhead unnecessarily. Values above 240 seconds delay detection of stale sessions.
  7. Set Session Timer Behavior to Strict. This ensures that if the remote endpoint does not support Session Timers, the call is rejected or falls back to a negotiated lower bound, rather than silently disabling the feature.

Architectural Reasoning: Setting the interval to 180 seconds balances overhead and detection speed. A re-INVITE or UPDATE is exchanged every 180 seconds. If a network drop occurs, the next scheduled refresh will fail. The SIP stack will retry according to its retry policy. If the retry fails, the session is torn down. This prevents the session from lingering indefinitely.

The Trap: Configuring Session Timers on a trunk that connects to a carrier that does not support RFC 4028. If the carrier does not support Session Timers, it may ignore the Session-Expires header or respond with a 420 Bad Extension. In Genesys Cloud, this can result in call setup failures. Always verify carrier support before enabling strict mode. If the carrier does not support it, enable Session Timer Support but set Behavior to Optional and monitor for 420 errors.

NICE CXone Configuration

In NICE CXone, Session Timer configuration is less granular and often depends on the specific SIP Interconnect profile.

  1. Navigate to Admin > Telephony > Trunk Management.
  2. Select the target trunk.
  3. Locate the SIP Settings or Advanced SIP Options tab.
  4. Enable Session Timer.
  5. Set the Timer Value to 180 seconds.
  6. Ensure Require Session Timer is unchecked unless the carrier explicitly mandates it.

Architectural Reasoning: NICE CXone’s implementation is more passive. It will include the Session-Expires header in INVITEs if enabled. If the carrier responds with a Session-Expires header, NICE CXone will honor it and send periodic refreshes. If the carrier does not include it, NICE CXone will not enforce timers. This is safer for interoperability but less robust for stale session detection. You must rely on TCP Keepalives and Media Detection as secondary mechanisms.

3. Configuring TCP Keepalives and SIP OPTIONS

TCP Keepalives are operating system-level mechanisms that send empty packets to verify the TCP connection is alive. SIP OPTIONS requests are application-level pings.

Genesys Cloud CX Configuration

Genesys Cloud uses TCP Keepalives on the SIP signaling channel.

  1. Navigate to Admin > Telephony > Trunks.
  2. Select the target trunk and click Edit.
  3. Expand the Advanced Settings section.
  4. Locate the TCP Keepalive configuration block.
  5. Set Keepalive Interval to 30 seconds.
  6. Set Keepalive Timeout to 90 seconds.

Architectural Reasoning: A 30-second interval ensures that NAT devices and firewalls do not expire the TCP session. Most enterprise firewalls have a default TCP idle timeout of 5-10 minutes. However, carrier-side firewalls may have shorter timeouts. A 30-second interval is aggressive enough to keep the connection alive without generating excessive traffic. The 90-second timeout ensures that if three consecutive keepalives are missed, the connection is considered dead.

The Trap: Setting the Keepalive Interval too low (e.g., 5 seconds). This generates unnecessary network traffic and can trigger rate-limiting on carrier firewalls, leading to connection resets. Setting it too high (e.g., 120 seconds) risks the carrier firewall dropping the connection before the next keepalive is sent.

NICE CXone Configuration

NICE CXone does not expose TCP Keepalive settings directly in the UI for all trunk types. For SIP Interconnects, you must configure this in the SIP Profile.

  1. Navigate to Admin > Telephony > SIP Profiles.
  2. Select the profile associated with the trunk.
  3. Locate Keepalive Settings.
  4. Set Keepalive Interval to 30 seconds.
  5. Set Keepalive Count to 3.

Architectural Reasoning: The Keepalive Count determines how many missed keepalives are tolerated before the connection is declared dead. A count of 3 with a 30-second interval results in a 90-second detection window. This aligns with the Session Timer refresh rate, providing a consistent detection window.

4. Implementing Media Health Checks (RTCP)

Signaling health does not guarantee media health. Orphaned sessions can have active signaling but dead media. RTCP (RTP Control Protocol) provides feedback on media quality and can detect media silence.

Genesys Cloud CX Configuration

Genesys Cloud enables RTCP by default. However, you can tune the behavior.

  1. Navigate to Admin > Telephony > Trunks.
  2. Select the target trunk and click Edit.
  3. Expand the Media Settings section.
  4. Ensure RTCP Support is Enabled.
  5. Set RTCP Interval to 5 seconds.

Architectural Reasoning: RTCP packets are sent every 5 seconds by default. This provides frequent feedback on packet loss, jitter, and delay. Genesys Cloud uses RTCP to detect one-way audio and media drops. If RTCP packets stop flowing, Genesys Cloud can tear down the session.

The Trap: Disabling RTCP to reduce bandwidth. RTCP overhead is minimal (less than 1% of total bandwidth). Disabling it removes the primary mechanism for detecting media health. Always keep RTCP enabled.

NICE CXone Configuration

NICE CXone also enables RTCP by default.

  1. Navigate to Admin > Telephony > SIP Profiles.
  2. Select the profile associated with the trunk.
  3. Locate Media Settings.
  4. Ensure RTCP is Enabled.
  5. Set RTCP Interval to 5 seconds.

5. Validating Configuration with API Payloads

To ensure consistency across environments, use the Genesys Cloud API to validate and deploy trunk configurations.

Genesys Cloud Trunk Update Payload

PATCH /api/v2/telephony/providers/edges/{edgeId}/trunks/{trunkId}
Content-Type: application/json

{
  "name": "CarrierTrunk-Prod",
  "description": "Production Trunk with Session Timers",
  "transport": "TCP",
  "sessionTimer": {
    "enabled": true,
    "interval": 180,
    "behavior": "strict"
  },
  "tcpKeepalive": {
    "enabled": true,
    "interval": 30,
    "timeout": 90
  },
  "mediaSettings": {
    "rtcpEnabled": true,
    "rtcpInterval": 5
  }
}

Architectural Reasoning: Using the API ensures that the configuration is version-controlled and reproducible. The PATCH method allows you to update only the relevant fields without overwriting the entire trunk configuration.

NICE CXone Trunk Configuration Validation

NICE CXone does not provide a direct API for updating trunk timer settings. You must validate the configuration via the UI or by analyzing SIP traces.

Validation, Edge Cases & Troubleshooting

Edge Case 1: Carrier Rejects Session Timer Negotiation

The Failure Condition: Calls fail to connect with a 420 Bad Extension error.

The Root Cause: The carrier gateway does not support RFC 4028 or has it disabled. The Genesys Cloud trunk is configured with behavior: strict, which rejects calls if the remote endpoint does not support Session Timers.

The Solution:

  1. Check the SIP trace for the 420 error.
  2. Update the trunk configuration to set behavior to optional.
  3. Alternatively, contact the carrier to enable Session Timer support.

Code Snippet:

{
  "sessionTimer": {
    "enabled": true,
    "interval": 180,
    "behavior": "optional"
  }
}

Edge Case 2: False Positive Session Tearing Down

The Failure Condition: Active calls are dropped after 180 seconds of silence.

The Root Cause: The carrier gateway is not sending the required re-INVITE or UPDATE requests to refresh the Session Timer. This can happen if the carrier’s SIP stack is misconfigured or if there is a network partition that blocks SIP traffic but allows media.

The Solution:

  1. Analyze SIP traces to confirm the absence of re-INVITE/UPDATE messages.
  2. Increase the sessionTimer.interval to 240 seconds to provide more tolerance for network latency.
  3. If the carrier cannot fix the issue, disable Session Timers and rely on TCP Keepalives and RTCP for session detection.

Edge Case 3: TCP Keepalive Storm

The Failure Condition: High CPU usage on the CCaaS edge and carrier gateway.

The Root Cause: The TCP Keepalive interval is set too low (e.g., 5 seconds) across thousands of concurrent channels. The volume of keepalive packets overwhelms the network interface cards.

The Solution:

  1. Increase the tcpKeepalive.interval to 30 seconds.
  2. Monitor CPU and network utilization on the edge devices.
  3. Ensure that the carrier firewall does not rate-limit TCP keepalives.

Official References