Serverless computing, explained
Servers still exist. The name means the cloud provider manages them behind the service instead of asking your team to operate them directly.
Serverless computing runs application code without requiring your team to provision or maintain the underlying servers, and the service can scale with demand.
The provider manages the capacity
Your team supplies code or an application service. The provider decides which computing resources run it and adds capacity when more work arrives.
Some serverless services run small functions after a trigger. Others run larger application components. In both cases, the managed infrastructure is the part that makes the approach serverless.
Less server work does not remove ownership
Serverless computing can reduce work such as choosing machine sizes or keeping idle servers ready. It does not decide who may access a record or whether an approval rule is correct.
Your business still owns the application logic, data choices, and checks that prove the result works. It should also understand the provider's limits and how the application could move if its needs change.
Serverless computing terms
- Function
- A small piece of code that performs one defined job when it is triggered.
- Trigger
- An event or request that tells a serverless service to run code.
- Automatic scaling
- The provider increases or reduces computing capacity in response to demand.
Where serverless fits
Uneven demand
Automatic scaling can suit work that is busy at some times and idle at others.
Defined jobs
A function can handle a focused task such as processing a submitted form.
Smaller operations load
The provider manages the servers that allocate capacity for the code.
Questions about serverless computing
Yes. The provider operates the servers and presents a managed service to your team. You do not manage those machines directly.
No. With serverless computing, a team supplies code or application logic. With software as a service, the provider delivers a complete application ready to use.
It may be a poor fit when the workload needs a fixed environment, requires control the service does not offer, or would be difficult to move away from the provider's specific features.
Related glossary topics
Focus on the tool your team needs
Caffeine apps run on the Internet Computer, a public network built for this.