UK What an edge data centre does
An edge data centre, written edge data center in US usage, is a smaller data centre placed close to the users, devices and machines it serves. It runs applications and stores data locally, so requests do not have to travel to a distant cloud region and back.
The aim is to shorten the path. Every network hop and every few hundred kilometres of fibre add delay, and sending raw sensor or video data to a central site costs bandwidth. An edge data centre handles the time-critical and data-heavy work nearby, then sends summaries, results or backups to the core data centre or cloud.
There is no fixed size. An edge data centre can be a single sealed cabinet at a factory or cell tower, a modular room in a car park, or a regional colocation building serving a city. What they share is location and purpose: close to the source, and run mostly unstaffed.
Edge, core and cloud compared
Edge data centres sit in a chain that runs from the devices themselves to the hyperscale cloud. Each layer has a different job.
| Feature | Device edge | Edge data centre | Core or cloud data centre |
|---|---|---|---|
| Location | On the machine, vehicle or pole | On site or in the same town | Regional or national hub |
| Scale | One computer | One cabinet to a few hundred racks | Thousands of racks |
| Typical latency to users | Under a millisecond | Typically a few milliseconds | Often tens of milliseconds |
| Typical hardware | Embedded and edge AI computers | Short-depth and rack servers, UPS | Rack and blade servers at scale |
| Staffing | None | Mostly unstaffed, remote managed | Staffed around the clock |
Latency figures depend on distance and network design, so treat them as typical. The point is the order of magnitude: an edge site removes most of the distance from the round trip.
Types of edge data centre
Edge sites are usually grouped by size and by who runs them.
-
//
Micro data centre. One or two self-contained racks with built-in cooling, UPS and fire protection, placed in a factory, shop or plant room.
-
//
Modular or containerised. Prefabricated rooms or converted shipping containers delivered complete and installed on a pad, often at telecom or utility sites.
-
//
On-premises edge. A server room or cabinet run by the site owner, close to the production lines, depot or hospital it serves.
-
//
Telecom edge (MEC). Compute placed in mobile operators’ networks for multi-access edge computing, so 5G applications run near the radio.
-
//
Regional colocation. Smaller data centres in second-tier cities that rent rack space close to local users and networks.
What is inside an edge data centre
The equipment is the same family as a large data centre, scaled down and hardened for sites without full-time staff. Racks hold servers, network switches and storage. Power arrives through a UPS and often a generator connection, and cooling ranges from a cabinet air conditioner to the building’s own system.
Many edge sites have tight space and less controlled conditions than a purpose-built hall, so short-depth chassis, DC power and wider temperature ratings are common. Our rack mount servers include 1U systems with redundant AC or DC power and a 2U fanless server rated for wide temperatures. For AI work on many video streams, edge AI servers add several high-power GPUs. Our guide to AMD EPYC Embedded for edge servers covers processor choice for these sites.
Remote management is the other essential. With nobody on site, every server needs out-of-band management so it can be restarted or reinstalled from a distance.
What edge data centres are used for
The common thread is work that cannot wait for, or cannot afford, a trip to the cloud. Video analytics is a large share: a city, port or rail network with hundreds of cameras analyses footage locally and sends only events and clips onwards. Factories run quality inspection and machine monitoring close to the lines. Telecom operators run 5G network functions and low-latency services at the edge, and content delivery networks cache video and downloads near viewers.
Data rules matter too. Keeping data in the country, or on the operator’s own site, can be simpler when it is processed locally. Resilience is a further reason: a site that processes its own data keeps working through a wide-area outage and catches up when the link returns.
Quick answers on edge data centres
Edge data centres: common questions
What is the difference between edge computing and an edge data centre?
One is the approach, the other a place. Edge computing is the practice of processing data near its source. An edge data centre is one place that processing happens; it can also run on a single edge computer on a machine or vehicle.
How big is an edge data centre?
Anything from one cabinet up. Micro data centres are one or two racks. Modular sites hold tens of racks, and regional colocation edge sites can reach a few hundred, still small next to hyperscale halls.
Does an edge data centre replace the cloud?
No, it works alongside it. The edge handles time-critical and data-heavy processing, while the cloud or core data centre handles training, long-term storage and central services.
Why use an edge data centre?
For speed, bandwidth and control. Local processing cuts response times, reduces the data sent over wide-area links and keeps sensitive data on site or in the country.
What is a micro data centre?
A self-contained data centre in one or two racks. It combines servers, network, UPS, cooling and monitoring in a sealed enclosure that can be placed in a factory, shop or plant room.
We build edge and rack servers for these sites with the processors, memory, GPUs, storage and operating system fitted as standard, and validate each one before it ships. Tell us your rack space, power, conditions and workload, and we will specify the servers to match.