Ring Topology and Fast Failover: Building a Video Network That Stays Up

"The cameras went offline for forty minutes" is rarely a camera problem. It is usually a single uplink, a single power supply, or a single switch — one component with no alternative path. Redundancy in a video network is not about buying twice as much hardware; it is about making sure no single cable or box can take the whole site down.

1. Understand what a ring buys you

In a ring, switches are cabled in a loop and one link is logically blocked. If a link or a switch fails, the blocked link opens and traffic takes the other way round. The site stays up; the only visible effect should be a brief re-convergence.

Protection	Typical recovery	Where it fits
No protection (star)	minutes to hours (manual)	single cabinet, non-critical sites
RSTP / MSTP	roughly a second or two	general purpose, mixed vendors
ERPS (G.8032) ring	well under 200 ms, typically far less	video networks where a visible gap is unacceptable
Redundant power + dual uplink	no gap if only one side fails	core cabinets, head-end

2. Ring design rules that actually matter

One ring master per ring; two masters fight and the ring oscillates.
Keep rings small — six to eight nodes is a comfortable ceiling, not a target.
Every node in the ring must support the same protocol; mixed ERPS and RSTP needs a documented boundary.
Ring ports should not also carry office or guest traffic.
Test the failover by pulling a cable during commissioning — with cameras recording, not on a bench.

3. Redundancy beyond the ring

A ring protects the path, not the power and not the recorder. The failures that still take a site down are usually:

One PSU in the cabinet — dual power input, or a UPS with monitored shutdown.
One recorder — a second NVR or edge recording on cameras.
One uplink to the head-end — a second physical path, ideally on different fibre.
One switch model in stock and no spare — keep one cold spare per site class.

4. Industrial sites raise the bar

On a plant floor or an outdoor cabinet the environment fails before the network does: 60 °C in summer, condensation in winter, vibration from machinery, and surge on long copper runs. A commercial switch in that cabinet will pass commissioning and fail in the second summer.

For those sites use industrial switches with a wide temperature rating and DIN-rail mounting, such as the RIS2000-8GP-2FM or the JIS2500-8PM-4GF-2SFP+ with fibre uplinks for the ring.

5. When redundancy is not worth it

A single-cabinet shop with eight cameras does not need a ring; it needs a UPS and a documented spare switch. Redundancy pays where downtime has a cost: 24-hour retail, logistics yards, utilities, remote sites where a callout takes half a day to reach.

Tell us the site layout and the acceptable downtime and we will propose a topology — ring or star — with the port count that matches. Get a quote.

FAQ

How fast does a ring actually recover?

ERPS-class rings recover well under 200 ms, often in tens of milliseconds; RSTP typically takes a second or two. Both are fast enough that recordings usually show no visible gap.

Can I build a ring with unmanaged switches?

No. A ring needs a managed protocol — ERPS or RSTP — to break the loop logically. Unmanaged switches in a physical loop will flood the network and take the site down.

Do I need a ring on a small site?

Usually not. A UPS, a spare switch and a documented configuration will prevent more outages than a ring on an eight-camera site.

What else fails besides the cable?

Power supplies, the recorder, and thermal limits in outdoor cabinets. A ring protects the path only — the rest needs its own answer.

Detailed technical specifications are provided in English.

← Learning