Edge computing or cloud computing: what is the difference?
Edge processes data close to the source, cloud centrally; many applications combine both for speed and scale.
Cloud computing processes data in large, central data centres. Edge computing does so closer to the source or the user, for example on a local server or in a nearby node. Edge reduces latency and cuts data traffic, while the cloud offers scale and central management. Many setups use both.
Who is it for: Developers, DevOps engineers and technical buyers who decide where processing takes place in their architecture.
What are cloud and edge?
In cloud computing, processing and storage run in large data centres that you use remotely. You benefit from scale, many services and central management. The drawback is the distance to sources and users, which adds latency.
Edge computing moves part of the processing to the edge of the network: a device, a local server in a factory or shop, or a node close to users. The aim is to process data before it travels to a central location.
How do they differ?
The main differences are location, scale and management. The cloud is easy to scale up and to manage centrally. Edge locations are more numerous, smaller and spread out, which makes managing software, security and updates harder.
Edge can keep processing when the connection is poor or drops. A machine in a factory can keep working if the connection to the cloud drops. A cloud application is more dependent on the network connection there.
When does edge suit?
Edge suits where response time is critical, where large amounts of raw data are pre-processed locally or where data may not leave a location.
- Real-time control and measurement systems with low latency requirements
- Pre-processing of image or sensor data to reduce traffic to the cloud
- Locations with variable or limited connectivity
- Data that must stay local for legal reasons
How do they work together?
In most designs, edge does the fast, local work and the cloud does the heavy, shared work: storage, training of models, reporting and management of all edge locations. Plan how you roll out software to many locations, how you keep secrets safe and what happens if the connection fails. A CDN is a simple form of edge for static content.
Cloud and edge compared
| Aspect | Cloud | Edge |
|---|---|---|
| Location | Central data centres | Close to the source or user |
| Latency | Higher because of distance | Lower because of proximity |
| Scale | Easy to scale up | Limited per location |
| Management | Central | Spread across many locations |
| On loss of connection | Dependent on the network | Can keep working locally |
Who is it not ideal for?
- Edge suits less when latency is not a problem; the extra management burden of many locations then yields little.
- Edge suits less for heavy, fluctuating compute loads that benefit from the elastic scale of the cloud.
- Teams without the resources for security and updates at many dispersed locations are better off choosing a central setup.
Frequently asked questions
Is a CDN edge computing?
A CDN is a form of edge for delivering content. Edge computing goes further and also runs logic at the edge.
Is edge more secure?
Not necessarily. Data stays local, but physical access to devices and remote patching bring their own risks.
What is fog computing?
A term for an intermediate layer between device and cloud, in practice a form of edge.
How do I update software at many edge locations?
With automated rollout, version control of the configuration and gradual rollout, so that a mistake does not hit everywhere at once.
