IPv6 and Dual-Stack Voice: Where an Enterprise SIP Network Breaks First
Most organisations enable IPv6 by publishing address records and turning the protocol on at the edge. Voice reacts differently, because two of its protocols are connectionless and hold no session state to fall back on. SIP signalling travels over UDP and RTP media travels over UDP, and neither renegotiates mid-call when the address family turns out to be wrong. The first symptom is usually a trunk that stops registering after a resolver restart.
Name Resolution Changes Before the Network Does
The failure is intermittent for a reason that makes it hard to pin down. Internal SIP servers on the same platform may be reachable over both address families, so registration and internal call setup continue while the path to the carrier does not. A test made with the resolver cache flushed will look broken while a test made from a cached desktop looks healthy, and the difference between them tells you what you were actually testing.
The useful diagnostic is to interrogate the resolver directly for the transport host, asking for the service record and the address records beneath it, then compare the answer set against the boundary route table. A record that resolves correctly but has no route back is a routing gap, and it presents as a registration timeout rather than a resolution failure.
| Layer | Dual-stack expectation | What actually happens |
|---|---|---|
| Address resolution | Both families returned, resolver prefers one | Preferred family changes after a cache flush |
| DNS64 and NAT64 | Translation is transparent to the client | Synthesised addresses expire mid-session |
| SIP Via and Contact | Address rewritten consistently | One field rewritten, the peer answers to a dead address |
| RTP source validation | Symmetric handling at both ends | One-way audio mistaken for a codec fault |
| Registration | Independent of resolver preference | Trunk stops registering after a resolver restart |
Translation Breaks the Dialogue, Not Just the Path
In a network with legacy-only uplinks, DNS64 synthesises address records for hosts that have only one, and a NAT64 gateway translates the resulting traffic. For web traffic this is transparent. For SIP it is not, because the address the far end learns in the Via and Contact headers is a translation artefact that stops being valid during the session. The signalling may look entirely correct in a capture, because it was correct when it was sent.
Media is worse. RTP is sent to the address and port the far end advertised, and a translator or stateful firewall will rewrite that address in one direction only unless symmetric handling is configured at both ends. The result is a call that rings, is answered, and carries one-way audio. The control channel travels separately, and its absence is easily mistaken for a codec fault.
The Trunk Is the Weak Link, Not the Handset
Endpoints are usually the easy part. A handset negotiates address families through the mechanism the platform implements and uses whatever the platform directs. The trunk and the inter-site links are where the assumptions live, particularly where a carrier publishes only one family and a session border controller is expected to cope with both. Ask the carrier for their position in writing before the change.
- Map which device resolves which name, rather than assuming the desktop does
- Query the resolver for the service and address records before blaming the trunk
- Confirm every preferred-family answer has a return route, not only a forward one
- Check whether translation rewrites Via and Contact consistently
- Verify symmetric RTP handling at both ends before blaming the codec
- Test with the resolver cache flushed, since cached answers hide the fault
- Keep a working legacy leg in place until the new leg has run a full week
A Migration Order That Reduces the Blast Radius
The order that works is unglamorous: resolve, then route, then advertise. Keep the published answer set unchanged while the routing underneath is proven, so the failover path keeps the same shape as the primary path. Enable the new family on the internal network first, where the addressing is under your control. Then bring up the carrier-facing leg with the old leg still in service. Only then publish the records that change what the platform resolves first.
Above all, keep the ability to go back. A dual-stack deployment that cannot be reversed in an evening has not been tested, it has been committed.
Redstone Approach
Redstone platforms carry SIP signalling and media over the address families the deployment provides, and trunk profiles keep separate signalling and media targets so a working leg can be retained while a second one is proven. Extension registration does not depend on the resolver's preference order, which removes the most common source of intermittent failure. Dual-stack endpoints can be introduced site by site, so a fault is contained to the site that introduced it.
Redstone Systems, Inc. founded in Delaware, USA in December 2002, has been the ODM vendor for many well-known communications companies, serving the Southeast Asian market. In 2020, Redstone Systems will begin to return to the North America market with its self-developed brand.
Redstone has a complete product line of intelligent voice gateways, providing IP-PBXs, analog VoIP gateways (FXS/FXO), digital VoIP gateways (E1/T1), border appliances, and session boundary controllers (SBCs).
With advanced technology in digital signal processor (DSP), speech coding and speech processing, as well as efficient operational tools such as cloud remote management, auto provisioning, Redstone gateways are widely used in markets of enterprise communications, cloud communications, call centers, operators’ IMS/SIP trunks, bringing users friendly, efficient and reliable communication experience.
For more information about Redstone, You can follow us on Facebook, Linkedin, and Youtube to be the first to get the latest news.

Current Location :

