Telecoms glossary • G
Geo-redundancy
Two data centres in two cities, both live. The difference between a bad hour and a bad day when one of them goes dark.
What is geo-redundancy?in hosted telephony and cloud services
1. Geo-redundancy means running a service from two or more data centres in different places, so that losing one site does not take the service down. For a hosted phone system it means your extensions, numbers and call routing exist in both locations, and calls keep connecting from whichever site is still standing.
2. It is a property of the provider’s platform, not of your office. A geo-redundant phone system still goes quiet at your desk if the single line into your building fails.
What an uptime number actually buys
Providers quote uptime as a percentage because it sounds precise. Converted into time, the gap between the numbers stops being abstract. A year is 525,600 minutes, so the arithmetic is easy to check for yourself.
| SLA | Downtime a year | Downtime a 30-day month | What that means in practice |
|---|---|---|---|
| 99% | 87 h 36 m | 7 h 12 m | Most of a working day, every month. Not a business telephony figure. |
| 99.9% | 8 h 46 m | 43 m | The common floor. A full working day lost across the year. |
| 99.95% | 4 h 23 m | 22 m | A single-site platform run well. |
| 99.99% | 52 m | 4 m | Realistically needs more than one site. Ask which ones. |
The percentage on its own is a marketing figure until you know what it covers and what it pays. Read what the SLA excludes, and read what a breach is worth: most credits are a percentage of the monthly fee, which for a small business is a refund of a few hundred rand against a day of missed calls.
Redundancy inside one building is not geo-redundancy
Every serious data centre is already redundant internally, and providers describe that resilience in language that sounds like geo-redundancy without being it. Teraco, where most South African voice platforms are hosted, specifies 99,999% uptime per facility. That is a well-run building, and it is still one building.
- Site resilience
- Dual power feeds, generators, N+1 cooling, two uplinks to different carriers. Protects against a component failing. Does not protect against the site itself becoming unreachable.
- Backups
- A copy of the configuration, restorable somewhere else. Protects your data, not your dial tone, and restoring is measured in hours.
- Cold standby
- A second site that exists but is not carrying calls. Someone has to notice and switch it on, and registrations have to move.
- Geo-redundancy
- Two live sites, in different places, both able to carry your calls now. Losing one changes which site answers, not whether anyone answers.
Distance is the point and also the cost. Johannesburg to Cape Town is around 1,260 km, which is roughly 6 ms of one-way latency in fibre before any equipment is involved. That is a trivial price inside the 150 ms budget a call has to work with, and it is why keeping both sites inside South Africa costs you nothing audible.
Four questions, and what a straight answer sounds like
-
Where are the sites, and are both live?
Name the cities and the facilities. Two halls in the same campus are one site for this purpose. An answer that names a provider rather than a location has not answered the question.
-
Is failover automatic, and how long does it take?
Ask for the number in seconds or minutes, and ask what triggers it. A person watching a monitor is not automatic failover.
-
What happens to calls that are already up?
Established calls usually drop even on a well-designed failover; new calls should connect immediately. A provider claiming that live calls survive a site loss is describing something unusual, so ask them to explain how.
-
Do the numbers follow?
Your direct numbers are delivered by an upstream carrier. If those routes point at one site only, the platform can be redundant while your inbound calls are not.
Ask for the status page and the incident history too. A provider that publishes outages, including the ones that were their fault, is worth more than one that publishes a percentage.
The half that is yours
Geo-redundancy is the provider’s side of the problem. Yours is the single fibre into the building, the router it terminates on and the power feeding both. A platform in two cities does not help a business whose only failure mode is the trench outside.
Pair the question with failover routing, so that when your line drops the calls land on mobiles instead of ringing into an empty office. That combination, not the SLA percentage, is what keeps the phones answered.
About this entry
Definitions are written for South African business buyers and checked against primary sources – provider documentation, standards bodies and ICASA – not vendor marketing. Reviews are independent and sponsors are always disclosed. Read our editorial policy and scoring methodology.
Spotted something out of date, or think we have this wrong? Tell the editors and we will re-check it.