SCTP multi-homing for resilient primary and secondary paths
SCTP multi-homing lets an endpoint maintain several usable paths to the same peer while preserving one logical association. It is especially useful where an IP transport must survive a fibre cut, router failure or carrier outage without forcing a new connection.
Although operators often call one route primary and another secondary, SCTP itself does not define a fixed hierarchy. All validated paths can carry traffic, while congestion, loss, routing and application policy determine which one is preferred.
For Australian networks, the fallback must use genuinely different infrastructure, not merely a second IP address on the same link. Carrier paths across Sydney, Melbourne, Brisbane, Perth and Adelaide can still share exchanges or ducts.
The following guide covers address planning, path verification, failover testing and the security controls needed for SIGTRAN-based services. It treats “primary” and “secondary” as operational roles rather than protocol labels.
How multi-homing creates resilient paths
A multi-homed SCTP association can span IPv4 and IPv6 addresses, or several routed interfaces over one IP family. Each local and remote transport-address combination forms a path that SCTP can validate, monitor and use independently.
Protocol background and related transport material are available in the SS7 Training library. Understanding that distinction prevents administrators from waiting for a formal primary-path election that SCTP does not perform.
Designing primary and secondary routes in Australia
Address planning starts with stable, routable addresses on each endpoint. A common pattern pairs an IPv4 path with an IPv6 path, but both routes must have independent reachability and compatible firewall policies.
Australian IPv4 scarcity and the auDA-led IPv6 transition make dual-stack planning increasingly common. Yet a supposedly secondary IPv6 route can fail if the upstream provider or enterprise firewall treats it differently from IPv4. Confirm route advertisements, source-address selection and return-path filtering for both families.
Access availability also varies. An NBN Enterprise connection may use fibre, fixed wireless or another access product depending on the premises and location. A backup bound to the same access network offers little protection, so operators should map the physical path to exchanges and carriers before calling it a secondary route.
Path monitoring and path selection
SCTP sends heartbeats on each path and maintains path-specific failure state, while retransmission timers govern association traffic. Path TSNs can verify a specific address tuple, but routing policy, congestion and source-address selection still decide which valid path is used. Verify behaviour with traffic counters rather than infer it from configuration.
In Australia, verification should include the far west, Tasmania and regional industrial sites, where fixed-network alternatives may be constrained. A 4G or 5G service from a different mobile carrier can provide control-plane resilience, but it needs a private APN with suitable addressing, plus tested power and routing behaviour.
Comparing active-active and active-standby designs
The main operational decision is whether healthy paths share work or one remains reserved. Active-active distributes streams across validated paths, whereas active-standby reserves the secondary until the primary fails.
Active-standby simplifies policy and may suit regulated or latency-sensitive services, but a standing secondary can hide ageing circuits. Active-active improves utilisation and exposes both routes during normal operation, although congestion control and source-address selection become more involved.
| Model | Normal traffic | Primary failure response | Main trade-off |
|---|---|---|---|
| Active-active | Shared across healthy paths | Traffic shifts to remaining capacity | Complex policy and capacity planning |
| Active-standby | Primary only | Backup starts after detection | Idle capacity and hidden defects |
| Hybrid | Critical streams on primary, overflow elsewhere | Policy selects protected path | Requires explicit stream prioritisation |
Building and validating the configuration
Build the configuration in layers. Assign each local and remote address, add routes and firewall rules for SCTP over IP, and use ASCONF where peer addresses are added dynamically. Confirm that both interfaces pass return traffic and that load-balancing policy does not repeatedly move an association between paths.
Next, set heartbeat and retransmission limits against the required failure-detection time, and enable SCTP-AUTH where supported. Monitor path state, retransmissions, round-trip time and association resets; thresholds must tolerate normal carrier delay without exceeding the service’s outage tolerance.
Test with the primary link disconnected, its route withdrawn and its carrier unavailable. Confirm convergence, stream continuity, source addresses and recovery after restoration. Use a maintenance window because failover can still disrupt stateful media or transaction sessions.
Signalling transport and security considerations
SCTP often carries SIGTRAN rather than traditional circuit-switched SS7, so primary and secondary path design still intersects with signalling architecture. For legacy interworking, the CIC allocation strategies guide explains how circuit identifiers are assigned across network boundaries.
Protect each path independently with source-address filtering, rate limits, IPsec where appropriate and platform-supported authentication. Permit SCTP only between known peers where feasible, since stateless traffic can otherwise be used for reflection attacks. CRC32c provides error detection, not peer authentication.
A practical deployment pattern for Australian networks
A workable Australian design starts with one preferred carrier path and one secondary that crosses a different access network and, ideally, a different exchange. Keep both endpoints multi-homed, and verify that the secondary is usable under normal load rather than treating it as a paper backup.
Broader technical context from Amir Zhan-Tassov can help teams connect transport resilience with signalling design.
For ports, mines, utilities and emergency coordination, include a third option such as an enterprise 4G/5G circuit where carrier addressing permits. Australian operators understand the need to give a backup a fair go, but power, DNS, certificates, tunnel endpoints and SCTP association settings must survive a blackout independently.
Practical takeaway: document the primary and secondary paths as separate, testable services, then repeat failover tests after every carrier, routing or firewall change. If either path shares the same physical fate, it is not an independent route.
