Application Deployment and Security · Topic 1 of 8
Edge Computing and Cloud Deployment Models
Learn the benefits of edge computing and compare private cloud, public cloud, hybrid cloud and edge deployment models by control, cost, scale and latency.
- 4 min
- Edge computing processes data near its source, giving low latency and lower bandwidth use.
- A private cloud is used by one organisation and gives the most control.
- A public cloud is shared, rented on demand and scales quickly.
What it is
A deployment model describes where an application runs and who owns and operates the infrastructure under it.
The four models in the outline are private cloud, public cloud, hybrid cloud and edge. Edge computing deserves its own look, because it moves processing close to where data is created.
The right model depends on cost, control, security rules, scale and how quickly users need an answer.
How it works
Each model trades control against convenience, and edge trades distance against speed.
Edge computing
Edge computing places compute and storage near the devices or users that produce the data, instead of in a distant data centre.
The benefits are:
- Low latency. A short path gives a fast response, which matters for control systems and interactive use.
- Less bandwidth. Local processing sends only results or summaries over the wide area network.
- Works with poor links. A site can keep running during a slow or broken connection.
- Data stays local. Sensitive data can be processed on site, which helps with privacy and data residency rules.
- Scale. Work is spread over many sites instead of one large centre.
Examples include a factory floor that checks product quality on camera, a retail store that runs its own point of sale, and a network branch that runs applications on its own router.
Private cloud
A private cloud is infrastructure used by a single organisation. It may sit in the organisation's own data centre or be hosted for it.
It gives the most control over security, compliance and design. The organisation pays for the hardware and the staff, and it has to plan the capacity itself.
Public cloud
A public cloud is infrastructure that a provider shares among many customers and rents out on demand. Examples are the large providers' services for virtual machines, storage and databases.
It scales fast and has little upfront cost, because you pay for what you use. You give up some control, and you depend on the provider's design and prices.
Hybrid cloud
A hybrid cloud joins a private cloud, or on premises systems, with a public cloud. Workloads and data can move between them.
It lets you keep sensitive systems private while using the public cloud for extra capacity, which is called bursting. It is also more complex to connect, secure and manage.
Comparing the models
| Model | Who runs it | Control | Scale | Typical reason to choose it |
|---|---|---|---|---|
| Private cloud | One organisation | Highest | Limited by own hardware | Strict compliance or data rules |
| Public cloud | A provider, shared | Lower | Very high, on demand | Fast growth, low upfront cost |
| Hybrid cloud | Both | Mixed | Extends on premises capacity | Keep some data private, burst to the cloud |
| Edge | Close to the data source | Local | Many small sites | Low latency, local processing |
Edge and cloud together
Edge does not replace the cloud. A common design processes data at the edge for speed, then sends results to a cloud for storage, reporting and training of models.
Edge sites are harder to manage, because there are many of them and some are remote. Automation, containers and central management tools help to keep them consistent.
Choosing for an application
A few questions guide the choice.
- Does the data have location or privacy rules? Think private cloud or edge.
- Does demand change a lot? Think public cloud.
- Do users or machines need an answer within milliseconds? Think edge.
- Is there an existing data centre that must keep working? Think hybrid.
If a scenario says low latency or poor connectivity at a remote site, the answer is edge. If it says burst capacity and keep sensitive data in house, the answer is hybrid.
What to remember for the exam
Six facts cover this topic.
- Edge computing processes data near its source, giving low latency and lower bandwidth use.
- A private cloud is used by one organisation and gives the most control.
- A public cloud is shared, rented on demand and scales quickly.
- A hybrid cloud combines private and public, and workloads can move between them.
- Edge often works together with a central cloud.
- The choice depends on cost, control, compliance, scale and latency.
Common mistakes
Four mistakes are typical.
Thinking edge replaces the cloud. It adds local processing and usually still sends data to a central cloud.
Calling any use of the cloud hybrid. Hybrid means that private or on premises systems are joined with a public cloud.
Assuming private means on premises. A private cloud can be hosted by a provider, as long as it serves one organisation.
Ignoring management effort at the edge. Many sites mean many systems to update and secure.
Moving data to a public cloud does not remove the organisation's duty to protect it. Rules about where data may live still apply.
Check yourself
Work through these three questions.
Why does edge computing reduce latency?
Processing happens close to where the data is created, so the path is short and the response is fast.
What is the difference between private and hybrid cloud?
A private cloud serves one organisation. A hybrid cloud joins private or on premises systems with a public cloud.
When might a company choose a public cloud?
When demand changes quickly and it wants on demand scale with little upfront cost.
Topic 1 of 8 in Application Deployment and Security
Test yourself on Application Deployment and Security
The full Cisco CCNA 200-901 DEVASC question bank has 808 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Cisco. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.