LTE Failover: How to Ensure Connectivity for Your Critical Sites

· 8 min read
8 min read
Eziwan Team
Eziwan Team
IoT Infrastructure

An industrial site without connectivity is like an operator working blind. No SCADA monitoring, no remote access, no fault alerts. For critical sites—pumping stations, electrical substations, continuous production facilities—a 10-minute network outage can cost several thousand euros.

The solution: automatic LTE failover with Dual SIM.

Why a Single Operator Isn't Enough​

Even mobile networks go down. The most common causes are:

  • Operator maintenance: 2–4 hours, sometimes unannounced
  • Network congestion: industrial areas with peak traffic (7 a.m.–9 a.m., 5 p.m.–7 p.m.)
  • Local antenna issue: only one BTS antenna covers your area
  • Weather: lightning, ice on radio equipment
  • Construction: fiber cut, damaged relay

Statistically, a single carrier offers 99.5–99.7% network uptime. That seems like a lot, but it amounts to 13–26 hours of downtime per year.

With two separate operators, the combined uptime is 99.97%—which amounts to less than 2 hours and 30 minutes of cumulative downtime per year.

Dual SIM Architecture: How Failover Works​

The Eziwan Gateway has two physical SIM slots. The recommended configuration:

SIM1: Carrier A (Orange)   → Primary connection
SIM2: Carrier B (SFR) → Failover connection

Failover is based on a ping watchdog: the router continuously tests the connectivity of the active SIM by sending ICMP pings to defined targets (8.8.8.8, the MQTT broker, your central server, etc.). If X consecutive pings fail within a window of Y, the failover is triggered.

{
  "failover": {
    "enabled": true,
    "primary_sim": 1,
    "watchdog": {
      "targets": ["8.8.8.8", "1.1.1.1", "mqtt.eziwan.com"],
      "interval_sec": 10,
      "timeout_ms": 3000,
      "fail_threshold": 3,
      "success_threshold": 5
    },
    "cooldown_sec": 120,
    "notifications": {
      "webhook": "https://hooks.slack.com/...",
      "email": ["admin@site.fr"]
    }
  }
}

This configuration triggers a failover if 3 consecutive pings fail (30-second detection window), and switches back to the primary SIM if 5 successive pings succeed (50-second stability window before switching back).

The 4 Failover Modes​

Eziwan supports four different strategies depending on your needs:

SIM1 is always active. SIM2 is activated only if SIM1 fails. Advantage: controlled data costs (SIM2 does not use data when SIM1 is working properly).

Mode 2: Round-robin​

Traffic switches between SIM1 and SIM2 depending on signal quality. Advantage: It always uses the best available network.

Mode 3: Load Balancing​

Traffic is distributed across both SIM cards in parallel. This doubles the available bandwidth. Useful for websites with large amounts of data.

Mode 4: Active/Active with Priority​

Both SIM cards are active, with intelligent routing based on traffic type (critical traffic on SIM1, best-effort on SIM2).

Case Study: Drinking Water Pumping Station​

One of our clients—a water district that manages 23 pumping stations—had a recurring problem: Orange service outages (2–3 times a year per station) made monitoring impossible and required manual preventive rounds.

Deployed Configuration:

  • SIM1: Orange (best rural coverage in their area)
  • SIM2: SFR (separate network, independent antennas)
  • Watchdog: ping every 10 seconds, failover if 3 pings fail
  • Notification: webhook → Nagios monitoring system

Results over 8 months:

  • 7 failover events (compared to 0 monitoring sessions without failover)
  • Average failover time: 14 seconds
  • Total actual downtime: 0 (transparent failover for SCADA)
  • Manual checks avoided: 7

What You See on the Dashboard During a Failover​

[14:32:01] ⚠ SIM1 Orange: Signal degradation — RSSI -97 dBm
[14:32:11] ⚠ SIM1 Orange: 3/3 pings KO (8.8.8.8, timeout)
[14:32:12] → Basculement SIM2 SFR en cours...
[14:32:18] ✓ SIM2 SFR: Connected — RSSI -72 dBm — Latency 28 ms
[14:32:19] ✓ VPN tunnel restored
[14:32:19] ✓ MQTT Broker Available — Monitoring Resumed
[14:32:20] ✉ Notification sent: admin@syndicat-eaux.fr

[15:44:02] ✓ SIM1 Orange: Signal restored (5 out of 5 pings OK)
[15:44:03] → Retour SIM1 principale
[15:44:09] ✓ Main connectivity restored

Total downtime: 7 seconds (between the last KO ping and the VPN being restored on SIM2).

LTE Failover + Wired WAN: Multi-Technology Redundancy​

Dual SIM redundancy protects against mobile network outages. But a comprehensive high-availability architecture adds a third dimension: technology redundancy—LTE plus fixed-line (ADSL, VDSL, or fiber optics).

When Dual SIM Isn't Enough​

Dual SIM (two mobile carriers) does not protect against:

  • Loss of geographic coverage: A cell tower covering an entire site area may be out of service due to a storm or fire
  • Massive network congestion: During exceptional events (disasters, military exercises), both mobile networks may become overloaded simultaneously
  • Regulatory constraints: Certain sectors (OIV, nuclear) require backup systems that differ from the primary technology

Hybrid WAN Architecture (LTE + ADSL/Fiber)​

WAN primaire  :  ADSL/VDSL/fibre  →  |                          |
                                       | Routeur Eziwan dual-WAN  |  →  LAN site
WAN backup   :  LTE 4G SIM1      →  |  failover automatique     |
WAN emergency:  LTE 4G SIM2      →  |                          |

The Eziwan Gateway can combine a wired WAN (Ethernet WAN port) and two LTE SIM cards according to a failover hierarchy:

  • Level 1: ADSL/fiber (best latency, carrier SLA)
  • Level 2: LTE SIM1 (Orange — best national coverage)
  • Level 3: LTE SIM2 (SFR — separate network as a last resort)

Multi-WAN Monitoring​

The watchdog monitors all three interfaces simultaneously:

{
  "wan_failover": {
    "interfaces": [
      {"name": "eth0-wan", "type": "fiber", "priority": 1},
      {"name": "lte0-sim1", "type": "lte", "priority": 2},
      {"name": "lte0-sim2", "type": "lte", "priority": 3}
    ],
    "watchdog": {
      "targets": ["8.8.8.8", "mqtt.eziwan.com"],
      "interval_sec": 15,
      "fail_threshold": 3
    }
  }
}

Calculate the Actual Availability of Your Failover Architecture​

The availability of a system with failover is calculated using the following formula:

Combined availability = 1 - (SIM1 unavailability × SIM2 unavailability)

ConfigurationSIM1 AvailabilitySIM2 AvailabilityCombined AvailabilityAnnual Downtime
No failover (SIM1 only)99.5%—99.5%~44 h/year
Dual SIM (SIM1 + SIM2)99.5%99.5%99.9975%~13 min/year
Triple WAN (fiber + LTE×2)99.9%99.5%99.9999%~32 s/year

The difference between "Dual SIM" and "SIM-only" is staggering: from 44 hours to 13 minutes of cumulative annual downtime. This is the technical justification for failover at critical sites.

Be aware of correlations: These calculations assume independent outages. If both operators share the same physical infrastructure in your area (co-location in a radio cabinet), outages will be correlated, and actual availability will be lower than the theoretical calculation. This is why it is important to choose operators with geographically separate infrastructures.

Best Practice: Test Your Failover Regularly​

A failover that has never been tested is a failover that could fail at the worst possible moment. Eziwan lets you trigger a forced failover test from the dashboard:

  1. Dashboard → Device → Actions → "Test Failover"
  2. Eziwan forces a failover to SIM2 and measures the failover time
  3. You verify that SIM2 is operational
  4. Automatic return to SIM1

Recommendation: Test once a month on each critical site.

Frequently Asked Questions​

Is the 14-second failover time guaranteed? No—this is a result observed in a specific scenario. The failover time depends on: the watchdog configuration (number of failed pings required, interval), how quickly the modem can obtain an IP address on the secondary SIM, and the time it takes to reestablish the VPN tunnel. An aggressive configuration (3 pings × 10s = 30s detection time) results in failovers within 30–60 seconds. A conservative configuration (5 pings × 30s) takes 3–5 minutes.

Do you need to have two different SIM contracts with two different carriers? Yes—for true failover, both SIM cards must be with different carriers (Orange + SFR, or Orange + Bouygues). Two SIM cards from the same carrier will be affected by the same local cell tower outage. The benefit of dual SIM is having physically separate networks.

Does dual-SIM failover protect against all outages? It protects against carrier outages (BTS antenna failure, network maintenance, network congestion). It does not protect against: a power outage at the site (connect the router to a UPS), a hardware failure of the router itself (have a replacement router or a swap agreement in place), or a geographic area with no LTE coverage at all (rare in mainland France).

Does the VPN reconnect automatically after a failover? Yes. The Eziwan Gateway keeps the VPN tunnel (OpenVPN or WireGuard) active on the new connection. With IPSec/IKEv2 + MOBIKE, the tunnel even survives a change in the public IP address. With OpenVPN (persist-tun), reconnection occurs automatically within a few seconds.

How do I monitor failover events? The Eziwan dashboard displays a complete history of connectivity events by site. You can configure webhooks to your monitoring system (Nagios, Zabbix, PagerDuty) or email/SMS alerts for each failover event.

Further Reading​

Do you have sites where a network outage could have critical consequences? Let's talk about it — our technical team can help you design the right failover architecture for your infrastructure.

Additional Resources​