SIP trunk redundancy and failover design

See also

Sources

1. 1
   Applying the same CUBE options-keepalive profile to dial peers with different bind interfaces is not supported, and the feature is not supported in multi-VRF setups.

   Cisco Unified Border Element Configuration Guide - Cisco IOS XE 17.6 Onwards - SIP Trunk Monitoring · SIP Trunk Monitoring, Restrictions · Checked 2026-09-25
2. 2
   CUBE out-of-dialog OPTIONS ping can mark a SIP dial peer busyout on heartbeat failure, and a dial peer is busied out only when all of its destinations fail to respond.

   Cisco Unified Border Element Configuration Guide - Cisco IOS XE 17.6 Onwards - SIP Trunk Monitoring · SIP Trunk Monitoring, OOD OPTIONS ping overview and server-group behaviour · Checked 2026-09-25
3. 3
   CUBE OPTIONS keepalive defaults are an up-interval of 60 seconds, a down-interval of 30 seconds and a retry count of 5.

   Cisco Unified Border Element Configuration Guide - Cisco IOS XE 17.6 Onwards - SIP Trunk Monitoring · SIP Trunk Monitoring, options-keepalive parameters · Checked 2026-09-25
4. 4
   A CUBE SIP options-keepalive profile can be shared by a group of dial peers so that one set of OPTIONS pings is sent, and if that ping fails all associated dial peers are busied out.

   Cisco Unified Border Element Configuration Guide - Cisco IOS XE 17.6 Onwards - SIP Trunk Monitoring · SIP Trunk Monitoring, OOD OPTIONS ping group · Checked 2026-09-25
5. 5
   When a CUBE dial-peer target resolves by DNS SRV, CUBE establishes out-of-dialog keepalive sessions with each resolved host and can distribute calls across the ones that are active.

   Cisco Unified Border Element Configuration Guide - Cisco IOS XE 17.6 Onwards - SIP Trunk Monitoring · SIP Trunk Monitoring, DNS SRV section · Checked 2026-09-25
6. 6
   A Unified CM 8.6(1) SIP trunk supports up to 16 destination IP addresses, 16 FQDNs, or a single DNS SRV entry.

   Cisco Unified Communications Manager Trunks (Session Management Edition deployment guide 8.6(1)) · SIP trunk destination addresses section · Checked 2026-09-25
7. 7
   In Unified CM 8.6(1) a SIP trunk is in service when at least one node receives a response other than 408 or 503 from at least one destination address.

   Cisco Unified Communications Manager Trunks (Session Management Edition deployment guide 8.6(1)) · SIP OPTIONS Ping section · Checked 2026-09-25
8. 8
   In Unified CM 8.6(1) SIP OPTIONS Ping, a trunk destination is out of service if it does not respond to OPTIONS, returns 503 or 408, or a TCP connection cannot be established.

   Cisco Unified Communications Manager Trunks (Session Management Edition deployment guide 8.6(1)) · SIP OPTIONS Ping section · Checked 2026-09-25
9. 9
   CUCM SIPTrunkOOS reason codes distinguish request timeout (Local=1), inability of the local SIP stack to create a socket connection (Local=2) and DNS query failure (Local=3).

   Configure Options Ping Between CUCM and CUBE · Troubleshoot section · Checked 2026-09-25
10. 10
    Cisco's CUCM-to-CUBE OPTIONS ping technote says OPTIONS messages are exchanged every 60 seconds by default once the feature is enabled on the SIP profile.

    Configure Options Ping Between CUCM and CUBE · Configure / Verify sections · Checked 2026-09-25
11. 11
    The Unified CM 8.6(1) guide says multiple prioritized SIP trunks are often needed for failure scenarios and should be placed in route groups in a single route list associated with a route pattern.

    Cisco Unified Communications Manager Trunks (Session Management Edition deployment guide 8.6(1)) · Multiple trunks / high availability section · Checked 2026-09-25
12. 12
    In Unified CM 8.6(1), Run on All Active Unified CM Nodes creates a SIP trunk daemon instance on every call-processing subscriber, so trunk calls can be sent or received on any of them.

    Cisco Unified Communications Manager Trunks (Session Management Edition deployment guide 8.6(1)) · Run on All Active Unified CM Nodes section · Checked 2026-09-25
13. 13
    The Unified CM 8.6(1) guide suggests a DNS SRV trunk destination when SRV host prioritization or weighting is needed, when more than 16 addresses are needed, or when the far end requires SRV resolution.

    Cisco Unified Communications Manager Trunks (Session Management Edition deployment guide 8.6(1)) · DNS SRV destinations section · Checked 2026-09-25
14. 14
    Failover triggers are not uniform across hops: RFC 3263 fails over on 503, transport failure or timeout; Teams defaults to 408/503/504 plus a 10-second timer; Twilio origination uses a 4-second no-response rule with an exclusion list; CUCM OPTIONS ping treats 408/503 as down. End-to-end failover time therefore has to be measured per hop rather than assumed.inferred

    Set-CsOnlinePSTNGateway (MicrosoftTeams) · Parameters -FailoverResponseCodes and -FailoverTimeSeconds, read against the other sources listed in corroborated_by · Checked 2026-09-25
15. 15
    Enterprise-side SBC redundancy does not by itself protect inbound DIDs: when every enterprise SBC is unreachable, inbound calls survive only through carrier-side configuration such as multiple origination targets or a carrier disaster-recovery destination.inferred

    Elastic SIP Trunking · Origination SIP URIs; Disaster Recovery URL · Checked 2026-09-25
16. 16
    Without an OPTIONS keepalive marking a peer down in advance, an element that fails over only on transaction timeout can wait on the order of Timer B (32 s at default T1) before trying the next server, unless a platform-specific shorter timer applies.inferred

    RFC 3261 — SIP: Session Initiation Protocol · Section 17.1.1.2 combined with RFC 3263 section 4.3 · Checked 2026-09-25
17. 17
    An SBC that has not sent OPTIONS on the expected interval is demoted: it is not tried first, but it is tried again before a call failure is generated if the other SBCs in the route cannot deliver the call.

    Monitor Direct Routing · Monitoring availability of Session Border Controllers using SIP options messages, demotion example · Checked 2026-09-25
18. 18
    Setting -Enabled $false on an online PSTN gateway drains it: existing calls continue and new calls go to another SBC in the route if one exists. A newly created gateway defaults to disabled.

    Set-CsOnlinePSTNGateway (MicrosoftTeams) · Description; Parameters, -Enabled · Checked 2026-09-25
19. 19
    The default FailoverResponseCodes on a Teams online PSTN gateway are 408, 503 and 504, and setting the parameter overwrites the defaults rather than adding to them.

    Set-CsOnlinePSTNGateway (MicrosoftTeams) · Parameters, -FailoverResponseCodes · Checked 2026-09-25
20. 20
    FailoverTimeSeconds defaults to 10: outbound calls the gateway does not answer within that time are routed to the next available trunk, and dropped if there is none.

    Set-CsOnlinePSTNGateway (MicrosoftTeams) · Parameters, -FailoverTimeSeconds · Checked 2026-09-25
21. 21
    Microsoft instructs administrators to allow Direct Routing signaling to and from all Teams IP ranges (52.112.0.0/14 and 52.120.0.0/14 for IPv4), not only the addresses returned by DNS.

    Plan Direct Routing · SIP signaling: FQDNs, Important note · Checked 2026-09-25
22. 22
    When two or more SBCs in one Direct Routing voice route are healthy and equal, Microsoft distributes calls between them with a Fisher-Yates shuffle.

    Monitor Direct Routing · Monitoring availability of Session Border Controllers using SIP options messages, final paragraph · Checked 2026-09-25
23. 23
    GCC High uses sip.pstnhub.gov.teams.microsoft.us and DoD uses sip.pstnhub.dod.teams.microsoft.us, and Microsoft states that secondary and tertiary FQDNs are not required in these environments.

    Plan Direct Routing · SIP signaling: GCC High; SIP signaling: DoD · Checked 2026-09-25
24. 24
    For a forwarded or transferred inbound PSTN call, if the ingress SBC is also a candidate egress SBC, Direct Routing ignores its route priority and tries it first.

    Configure call routing for Direct Routing · Example 1, Note after the three-route summary table · Checked 2026-09-25
25. 25
    IPv6 Direct Routing SBCs use sip.ipv6, sip2.ipv6 and sip3.ipv6 .pstnhub.microsoft.com in the same primary, secondary and tertiary order.

    Plan Direct Routing · SIP signaling: FQDNs, 'FQDNs in case SBC uses IPv6' table · Checked 2026-09-25
26. 26
    Direct Routing considers an SBC healthy if it sent SIP OPTIONS within the last three minutes (three times the one-minute regular interval) at the moment a call is routed.

    Monitor Direct Routing · Monitoring availability of Session Border Controllers using SIP options messages · Checked 2026-09-25
27. 27
    Direct Routing health monitoring is based on OPTIONS sent by the SBC to Microsoft, and no tenant administrator action is needed to enable it.

    Monitor Direct Routing · Monitoring availability of Session Border Controllers using SIP options messages, first paragraph · Checked 2026-09-25
28. 28
    For Microsoft 365, Office 365 and GCC, Direct Routing SBCs are configured with three SIP proxy FQDNs in priority order: sip.pstnhub.microsoft.com (primary), sip2.pstnhub.microsoft.com (secondary) and sip3.pstnhub.microsoft.com (tertiary).

    Plan Direct Routing · SIP signaling: FQDNs table · Checked 2026-09-25
29. 29
    Microsoft states that if the primary Direct Routing region becomes unavailable, traffic is redirected to the secondary and then the tertiary region without administrator intervention.

    Plan Direct Routing · SIP signaling failover · Checked 2026-09-25
30. 30
    In Direct Routing, SBCs inside one voice route are tried in random order, and if none in the higher-priority route is available, the next lower-priority route matching the same pattern is tried; if no SBC is available the call is dropped.

    Configure call routing for Direct Routing · Example 1: Voice routing with one PSTN usage, Call Flow 2 and following note · Checked 2026-09-25
31. 31
    SendSipOptions defaults to True; if disabled, the SBC is excluded from the Direct Routing monitoring and alerting system, and Microsoft highly recommends leaving SIP options enabled.

    Set-CsOnlinePSTNGateway (MicrosoftTeams) · Parameters, -SendSipOptions · Checked 2026-09-25
32. 32
    Microsoft recommends configuring redundant SIP connectivity and testing failover between SIP connection points before production rollout.

    Plan Direct Routing · Best practices; Verify your deployment · Checked 2026-09-25
33. 33
    PSTN usages in a Teams voice routing policy are evaluated in order, and once a match is found in one usage the later usages are never evaluated, so a catch-all usage placed first bypasses backup routes defined in a later usage.

    Configure call routing for Direct Routing · Example 2: Voice routing with multiple PSTN usages, Note under routing table · Checked 2026-09-25
34. 34
    RFC 3261 defines OPTIONS as a request for querying capabilities that may be sent outside a dialog, and it is the method platforms reuse as a trunk keepalive.

    RFC 3261 — SIP: Session Initiation Protocol · Section 11 Querying for Capabilities · Checked 2026-09-25
35. 35
    RFC 3261 sets INVITE client transaction Timer B to 64*T1, and with the default T1 of 500 ms that is 32 seconds.

    RFC 3261 — SIP: Session Initiation Protocol · Section 17.1.1.2 Formal Description; Table 4 timer values · Checked 2026-09-25
36. 36
    Under RFC 3263 a SIP URI whose host is a domain name with an explicit port is resolved with A/AAAA records only, so SRV priority and weight are not consulted for that URI.

    RFC 3263: Session Initiation Protocol (SIP): Locating SIP Servers · Section 4.2 Determining Port and IP Address · Checked 2026-09-25
37. 37
    Under RFC 3263 a SIP client treats a request as failed toward a resolved server when the transaction layer reports a 503 response, a transport failure, or a transaction timeout (Timer B or Timer F firing).

    RFC 3263: Session Initiation Protocol (SIP): Locating SIP Servers · Section 4.3 Details of RFC 2782 Process · Checked 2026-09-25
38. 38
    RFC 3263 also lets a server fail over when sending responses, by looking up SRV records for the sent-by domain in the Via header and trying the next entry if the first fails.

    RFC 3263: Session Initiation Protocol (SIP): Locating SIP Servers · Section 5 Server Usage · Checked 2026-09-25
39. 39
    On such a failure RFC 3263 says the client SHOULD build a new request with a new Via branch ID and send it to the next element in the RFC 2782 ordered server list.

    RFC 3263: Session Initiation Protocol (SIP): Locating SIP Servers · Section 4.3 Details of RFC 2782 Process · Checked 2026-09-25
40. 40
    RFC 3263 transport selection uses an explicit transport parameter if present, otherwise a NAPTR lookup, then SRV lookups for supported transports if no NAPTR records exist, and finally defaults to UDP for sip or TCP for sips URIs.

    RFC 3263: Session Initiation Protocol (SIP): Locating SIP Servers · Section 4.1 Selecting a Transport Protocol · Checked 2026-09-25
41. 41
    In DNS SRV a client must attempt the reachable target with the lowest-numbered priority first; a lower number means more preferred.

    RFC 2782: A DNS RR for specifying the location of services (DNS SRV) · The format of the SRV RR, Priority field · Checked 2026-09-25
42. 42
    In DNS SRV the weight field distributes selection proportionally only among records that share the same priority; larger weights get a proportionally higher chance of selection.

    RFC 2782: A DNS RR for specifying the location of services (DNS SRV) · The format of the SRV RR, Weight field · Checked 2026-09-25
43. 43
    RFC 2782 says administrators should use weight 0 when there is no server selection to do, and that weight-0 records should have a very small chance of selection when weighted (>0) records are present at the same priority.

    RFC 2782: A DNS RR for specifying the location of services (DNS SRV) · The format of the SRV RR, Weight field · Checked 2026-09-25
44. 44
    A Twilio Elastic SIP trunk can have a Disaster Recovery URL pointing to a TwiML application that handles inbound calls (for example with an IVR) when the origination URIs are unavailable, billed at standard Twilio Voice rates.

    Elastic SIP Trunking · Disaster Recovery URL · Checked 2026-09-25
45. 45
    Twilio fails over to another origination URI when there is no SIP response after 4 seconds; responses 2xx, 400, 404, 405, 410, 416, 482, 484, 486 and 6xx do not trigger failover.

    Elastic SIP Trunking · Origination failover behaviour · Checked 2026-09-25
46. 46
    Twilio offers localized termination URIs per edge (for example example.pstn.ashburn.twilio.com), each resolving to several IPs across availability zones, and recommends failing over to another region during a regional outage.

    Elastic SIP Trunking · Localized termination URIs · Checked 2026-09-25
47. 47
    A Twilio Elastic SIP trunk can have up to 10 origination URIs.

    Elastic SIP Trunking · Origination SIP URIs · Checked 2026-09-25
48. 48
    Twilio Elastic SIP Trunking uses the origination SIP URI with the lowest priority value first and fails over to URIs of equal or higher value if the session fails; priority ranges 0 to 65535.

    Elastic SIP Trunking · Origination SIP URIs: priority and weight · Checked 2026-09-25
49. 49
    Among Twilio origination URIs with the same priority, weight (1 to 65535) sets proportional load distribution, with higher weight receiving more traffic.

    Elastic SIP Trunking · Origination SIP URIs: priority and weight · Checked 2026-09-25
50. 50
    Webex Calling certificate-based trunks can be configured with a DNS SRV address, which provides redundancy and load distribution toward the premises gateway.

    Configure trunks, route groups, and dial plans for Webex Calling · Trunks section, certificate-based trunk configuration · Checked 2026-09-25
51. 51
    A Webex Calling route group must contain at least one and at most 10 trunks, and can mix registration-based and certificate-based trunks.

    Configure trunks, route groups, and dial plans for Webex Calling · Route groups section · Checked 2026-09-25
52. 52
    In a Webex Calling route group, calls are distributed randomly among trunks of the same priority, and if the highest-priority trunk is unavailable the call is tried on a lower-priority trunk.

    Configure trunks, route groups, and dial plans for Webex Calling · Route groups section, priority setting · Checked 2026-09-25

Documents

Cite this page

WarmTransfer. (2026, September 25). SIP trunk redundancy and failover design. WarmTransfer. https://warmtransfer.net/knowledge/sip-trunk-redundancy-design

@misc{warmtransfer-sip-trunk-redundancy-design,
  title  = {SIP trunk redundancy and failover design},
  author = {{WarmTransfer}},
  year   = {2026},
  url    = {https://warmtransfer.net/knowledge/sip-trunk-redundancy-design},
  note   = {Verified 2026-09-25}
}