Petya Petrova - Fotolia
Admins should understand what serverless computing is and how it works, but function creation and deployment are mainly the domain of software developers; most admins are not involved with serverless computing infrastructures in their own company.
In an on-premises software deployment, an admin is responsible for provisioning compute and other resources needed to operate the workload, and then often handle the tasks of deploying that workload to the provisioned resources, such as a virtual machine and storage volume.
Agile software development practices use automation to provision and configure the deployment of each software build or iteration, so there is less direct admin involvement. In 2019, agile computing involves provisioning and deploying workloads into cloud resources, such as AWS EC2 and S3 instances.
With these cloud deployments, resources incur hourly costs, which is pricey and ineffective. Why should an organization pay for cloud resources if the workload is only used a relatively small fraction of the time?
Rather than an idle workload sucking up costly resources, cloud architects developed the idea of "functions." With a function, associated code can be loaded and executed in response to a trigger -- such as a real-life event or some other programmatic hook -- and is released as soon as the function is complete.
This way, users pay for functions based on the number of function calls and the total time that the function was actually executing during the billing cycle. Major providers of function as a service are AWS, Azure and Google.
Because developers do not need to provision or configure server resources to operate functions, the technology was dubbed "serverless computing." It’s important to note that functions need servers and other infrastructure; end users and developers just don’t see it.
Successful serverless architecture depends heavily on scale and automation. It is certainly possible for an enterprise with a private cloud to implement its own serverless architecture. A data center admin would be involved when an organization deploys a private cloud and wants functions on the in-house service menu.
Implementing a private cloud with functions as a service demands significant intellectual, technical and financial investment. In most cases, an organization with software designed to deploy functions benefits from the scale and automation already found in the cloud, and local data center admins must support and develop the local infrastructure for more conventional workloads.