Practical, testable failover steps you can run before your next high-volume campaign — designed for Malaysian PBX admins using SIP trunks.
Why this checklist matters: voice remains business‑critical in Malaysia even as data grows — outages translate to lost appointments, missed payments and frustrated customers. Use this checklist to make your SIP trunking resilient, measurable and auditable.
You’re the PBX admin who must keep agents live when campaigns run or when the office internet trips. SIP trunking gives you flexibility and lower costs, but it also introduces single points of failure if the design is opinionated around one ISP, one trunk registration, or a brittle DNS/TLS setup. This guide is focused on actionable configuration, testing, and operational controls you can finish in a maintenance window (or a coffee break) — with specific notes for Malaysian operators and examples that match common ITGTEL SIP Trunking plans.
Straight answer: instrument health checks at three layers — SIP signalling (OPTIONS/REGISTER), transport (ICMP/TCP), and application (synthetic test calls). Combine short OPTIONS polling (5–10s) with an SBC that interprets failure codes and an external monitor that alerts within 30 seconds. This multi‑layer detection avoids false positives and speeds failover decisions.
Technical detail: configure your PBX/SBC to send OPTIONS every 5–10 seconds to the carrier’s SIP proxy; reduce DNS TTLs for NAPTR/SRV/A records to 60–120s during high‑availability windows; terminate TLS at the SBC where possible so certificate issues don’t silently break registrations.
Further reading: 3CX — SIP Trunk Failover
Direct answer: these seven items cover detection, routing, redundancy, and verification — follow them in order and mark each as PASS/FAIL during your test. Done right, they reduce call loss during trunk or ISP outages from minutes to seconds.
Use at least two independent ISPs (different last‑mile providers) or an SD‑WAN with independent uplinks. Verify different AS paths with a traceroute to your SIP provider’s signalling IPs and confirm the two uplinks do not share local aggregation points.
Register your PBX/SBC to the primary carrier and a secondary carrier (or a secondary POP from the same carrier). If the carrier supports IP‑auth only, register the secondary via the carrier’s alternate IPs or use credentialed trunks as a backup.
Configure OPTIONS heartbeat to 5–10s, with a 3‑strike failure policy before failing over (e.g., 3 x 5s = 15s). Map SIP response codes: 404/408/503 → immediate failover; transient 486/480 → retry. Keep DNS TTLs low (60–120s) during cutover windows.
Use an SBC to centralise policy: prefer primary trunk, fallback to secondary, and finally route to a cloud PBX queue or voice blast service. The SBC should support translation, TLS/SRTP, SIP normalization, and per‑trunk call limits.
Plan DID failover so inbound numbers can be rehomed to alternate queues or IVRs within your PBX routing logic. Update your emergency numbers routing policy if your secondary trunk’s geographic origination differs.
When OPTIONS fail, trigger an automated test call from an external monitor (cloud probe) to confirm audio path. Alert both network and contact‑centre ops via SMS/WhatsApp and escalate if the probe fails three times in a row.
Simulate the primary trunk outage during off‑peak hours: disable the primary registration, measure switchover time, record missed calls, and file a post‑mortem. Keep results in your runbook and repeat quarterly.
Quick check: if your measured switchover time is >60 seconds, re-evaluate DNS TTLs, OPTIONS interval, and whether your SBC is the decision-maker — these three elements typically dominate latency.
Direct answer: busy call centres should prefer active‑active registrations (or registered SIP peers across multiple POPs) combined with an SBC that performs per‑call routing and dynamic load balancing. Active‑active removes the registration single‑point and lets carriers split traffic so failover is seamless.
For very high concurrency, combine active‑active trunks with call distribution logic on the PBX and persistent session handoff at the SBC. If active‑active isn’t available, use active‑hot standby with sub‑20s OPTIONS-based detection and pre‑warmed call queues on the backup.
Vendor reference on strategies: Cisco — High Availability and Network Design
Direct answer: DID continuity requires pre‑configured DID failover rules with your SIP provider and internal PBX mapping. The provider must support rehoming the DID (move the routing) to a secondary trunk or to a SIP URI you control; your PBX then maps that incoming URI to the correct queue or IVR.
Operational note: confirm with your Malaysia SIP trunk provider whether DDI rehoming is instant or needs a carrier update window. For high‑risk numbers (finance, bookings), pre‑configure voiced‑based fallback (IVR + recorded message) so calls do not drop during provider reconfiguration.
Direct answer: monitor three KPIs — registration health (OPTIONS pass rate), median failover time (seconds), and inbound call completion rate (%) during failovers. Target: OPTIONS pass rate 100% in normal state, median failover time ≤20s, and inbound call completion ≥95% during failover drills.
Add a synthetic call probe every 60s from an external cloud monitor located in Malaysia to validate audio path; log all failovers and keep a dashboard of last‑mile outages by ISP. Make SLA breach thresholds explicit for vendor escalation.
Direct answer: the usual problems are single-ISP dependency, long DNS TTLs, relying only on REGISTER failures to detect outages, and no synthetic tests that verify RTP media. Avoid these and you eliminate most common outages.
“In Malaysia the right strategy is ISP diversity + pre-warmed backup routes. Because fixed broadband penetration is not uniform, your PBX must assume at least one local link can fail.” — ITGTEL operations
Direct answer: ITG Telecommunications Sdn Bhd (ITGTEL) provides SIP Trunking plans with multi‑channel capacity, DDI number packs and managed failover options — including pre‑built routing rules and SBC provisioning — so you can offload the carrier side of failover to an MCMC‑licensed provider with local operation and support.
If you run high‑volume voice operations, consider plans with multiple voice channels and pre‑reserved DDI blocks (for example, ITGTEL’s Universe and Space plans) so inbound numbers can be remapped quickly during an incident. See our SIP Trunking service page and the Universe plan for plan details and provisioning options.
Internal resources: ITGTEL SIP Trunking service page, SIP Trunking – Universe – 12 Month, and our operational guide SIP Trunking – Cutting Telecom Costs for Businesses in Malaysia.
Compliance/PDPA note: when you re-route DDI or play recorded IVR messages during failover, ensure any customer personal data captured is handled per PDPA — document the change of data flow and retention for audits.
This is the one‑page snippet you can paste into an incident runbook and follow under pressure.
| Action | Command/Check | Target |
|---|---|---|
| 1. ISP diversity | traceroute to sip.pop1.example, sip.pop2.example | 2 independent AS paths |
| 2. OPTIONS heartbeat | OPTIONS interval 5–10s; 3 failures → failover | Failover ≤20s |
| 3. SBC routing | Primary → Secondary → CloudPBX | Clear precedence and limits |
| 4. DID rehoming | Verify carrier support; pre‑staged IVR | Inbound continuity ≥95% |
| 5. Synthetic test call | External probe call every 60s | Audio pass/fail logged |
| 6. Alerting | WhatsApp/SMS & email escalation | Ops alerted < 60s |
| 7. Drill | Quarterly simulation, record results | Post‑mortem within 48h |
Tip: include a pre‑approved message template (WhatsApp or SMS) for customer notification during long outages; this reduces support churn and preserves trust.
Data context: Malaysia fixed‑broadband penetration was 64.1 per 100 premises (Communication & Multimedia Dashboard via Department of Statistics Malaysia, as at June 2025). DOSM — TMP_MDE 2025 (source: MCMC Facts & Figures)
Run a light drill (registration failover) monthly and a full simulated outage quarterly. Document switchover times, missed calls, and vendor response times for each drill.
DNS SRV helps but is slow if TTLs are high and it doesn’t detect media/RTP problems. Combine SRV with OPTIONS polling, multiple registrations and an SBC decision layer for robust failover.
ITG Telecommunications Sdn Bhd waives the RM500 SIP registration fee on subscription for SIP Trunking plans. Contact our team for plan details and provisioning timelines.
Recommended engineering reading: 3CX – SIP Trunk Failover and Cisco – High Availability and Network Design.
