Just as serverless computing can dramatically improve the productivity of application developers, it can also be used to improve the productivity of infrastructure and operations (I&O) administrators responsible for managing public cloud services. Currently, the most common form of serverless computing is function Platform-as-a-Service (fPaaS), in which fine-grained units of custom application logic are packaged as functions that are executed when triggered by events. fPaaS functions can be triggered by events based on the changing state of infrastructure on infrastructure-as-service (IaaS) platforms, or in response to other operational activity. This allows fPaaS functions to be used for building powerful tools to automate operations on a cloud service.
However, to exploit fPaaS for management purposes, administrators will have to embrace a new operational approach based on event-driven architecture (EDA). This approach differs from traditional modes of operation, which typically involve some form of orchestration based on scripts, or newer specification-driven alternatives based on tools such as Ansible, Chef, Puppet, and Salt. fPaaS is ideally suited for applying EDA to automate infrastructure operations, but its use will require administrators to shift their perspective on the way they solve problems. In particular, they will have to look for key patterns identifying operations that are suitable for EDA-based automation. These patterns include statelessness, concurrency, idempotence, and the ability for events to record transitions in the state of systems (see Figure 1 below).
Figure 1: Principles of Event-driven Automation with Serverless Computing
By applying serverless computing to routine operational tasks, administrators can spend less time managing infrastructure, and turn their attention to other operational priorities such as security and monitoring. However, to accomplish this shift, administrators will have to reorient their perspective on the way they approach automation. This document (Gartner subscription required) provides more detail on how to apply serverless computing to help with infrastructure operations, and includes examples of using fPaaS to automate provisioning and management operations, and configure networking, reliability, and security.
On the surface, the rise of serverless computing would appear to obsolete the role of I&O teams. However, in practice, it is just one of a variety of tools that these teams should apply for automation when appropriate.