When a hospital IT manager evaluates internet, the first question is usually about speed — ‘How fast is it?’ The right question is ‘What happens when it fails?’ For hospitals, internet downtime isn’t an inconvenience — it’s a clinical event. PACS systems go offline. Electronic medical records become inaccessible. Telemedicine sessions drop. Lab results can’t be transmitted. This article explains why hospital internet architecture must start with failover, not speed.

What Happens When Hospital Internet Fails?

In a hospital without failover internet: Radiology’s PACS system loses connectivity, stopping image transfer from CT and MRI scanners to radiologists. The Hospital Information System (HIS) becomes unavailable, forcing manual paper-based processes. Telemedicine consultations drop mid-session. Cloud-based lab systems go offline. CCTV recording gaps appear. Administrative billing systems freeze. None of this is theoretical — it happens every time a single ISP has an outage.

How Often Do ISPs Go Down?

Even a premium ISP with a 99.9% uptime SLA has 8.7 hours of permitted downtime per year. That’s 8.7 hours of hospital systems potentially offline. In practice, most ISPs achieve 99.5–99.7% uptime — which means 13–26 hours of downtime per year. For a hospital running 24/7, that’s an unacceptable risk. Dual failover internet reduces your combined downtime to under 1 hour per year.

What Is Dual Last-Mile Failover and How Does It Work?

Dual last-mile failover means having two separate internet connections — each from a different provider, on a different physical path. If ISP A fails, traffic automatically routes through ISP B in under 30 seconds — without any manual action, without calling a helpdesk, without IT staff intervention. The failover device (a dual-WAN router or SD-WAN appliance) handles the switchover automatically. Staff don’t even notice it happened.

PACS and DICOM — Why Radiology Needs Dedicated Bandwidth

PACS (Picture Archiving and Communication System) transfers large DICOM files — a single CT scan can be 500 MB to 2 GB. If this traffic competes with general hospital internet usage, transfers slow to a crawl, radiologists wait, and diagnoses are delayed. The fix is dedicated PACS bandwidth on a separate QoS class — reserved bandwidth that general users cannot consume, even during peak hours.

Nurse Call Systems and Medical IoT

Modern nurse call systems, patient monitoring devices, and infusion pumps increasingly use IP-based communication over the hospital network. These systems are low-bandwidth but extremely latency-sensitive — a 500ms delay in a nurse call is unacceptable. Medical IoT devices need their own VLAN with QoS priority, isolated from general hospital internet to prevent interference.

Recommended Hospital Internet Architecture

For a hospital of any size, BTNL recommends: (1) Primary leased line sized for PACS + HIS + CCTV + general internet. (2) Backup leased line on a different physical route for automatic failover. (3) Dedicated QoS class for PACS/DICOM traffic. (4) Separate VLAN for CCTV with reserved bandwidth. (5) Patient WiFi on an isolated VLAN — completely separate from clinical systems. (6) Static IP for remote access to PACS and HIS. (7) 24/7 NOC monitoring with proactive failover testing.