At one time, the adoption of serverless platforms seemed beyond their comprehension. After some notable success stories, such as Capital One's “serverless first” strategy, the hype seemed to be dying down. This is in part because serverless has always struggled with a vague definition. Despite the lack of infrastructure and application servers, developers have had to have some understanding of operating within a cloud-native platform. According to Google search metrics, the evidence is clear. Around 2020, there was a significant drop in public interest in serverless platforms. However, in 2022, we began to see an interesting upward trend, and today, search metrics and survey data show that serverless is more popular than ever. Here are some reasons why serverless is trending again among customers:
As serverless became a path to developer productivity, a better definition emerged. The term “serverless” has always been a misnomer and has tended to mean different things among end users and vendors depending on the product and use case. Just as the cloud is someone else's computer, serverless is someone else's server. Today, the situation is much clearer. Serverless applications are software components that run in an environment that manages the underlying complexities of deployment, runtime, protocols, and process isolation so developers can focus on their code. Enterprise success stories provided proven, repeatable use case solutions. The initial hype around serverless was centered around rapid development cycles and backend use cases where serverless functions acted as glue between different cloud services. Many of these examples came from born-in-the-cloud early adopters and vendors leveraging Function as a Service (FaaS) inside their platforms and services. Since then, many more enterprise customers have embraced serverless. A growing ecosystem of ancillary services is giving rise to new use cases. The core use case for serverless is building lightweight, short-running, ephemeral functions. However, in recent years, serverless providers have expanded their ecosystem of data and integration services, such as serverless key-value stores, API frameworks, and serverless vector databases, enabling more diverse use cases.
A mature serverless ecosystem provides a new perspective on the market
In our previous Forrester Wave™ covering the serverless development space, we focused exclusively on FaaS platforms offered by hyperscale public cloud providers. While FaaS remains a core technology for serverless development, the ecosystem has expanded and become more focused on supporting use cases rather than adhering to a single platform approach. In our upcoming Landscape and Wave coverage of this market, we will incorporate the following emerging trends:
Versatile cloud services are replacing earlier serverless development use cases. Serverless computing has expanded beyond FaaS to refer to cloud-native development platforms that require little to no manual provisioning and offer autoscaling, consumption-based micro-billing, and scale to zero. Early functions were primarily used as an integration glue for use cases such as filtering, routing, batching, and event enrichment. These use cases are gone and replaced by flexible, turnkey cloud services built as part of a “compose-and-consume” development pattern. AI use cases are breathing new life into the serverless computing model. AI is a hot topic across most product categories and business units, and serverless computing is no exception. Given the pervasive impact of AI workloads in terms of power consumption, scarcity of GPU resources, and the inherent on-demand nature of generative AI prompts, the ephemeral serverless model has emerged as ideal for AI applications. WebAssembly (Wasm) has emerged as a powerful enabler for deployment-agnostic serverless platforms. Despite much of the buzz coming from vendors rather than end users, Wasm may be the future of serverless development platforms. Wasm has several built-in advantages: It has significantly faster cold start times, with demos showing it spin up in under a millisecond. It is also significantly more portable than other approaches, with the Wasm binary format being compatible with a wide range of browsers across multiple operating systems and CPU architectures, including Intel and ARM. Edge and serverless development will converge on distributed use cases. For some time, the infrastructure that service developers depended on, serverless or not, has existed in one of two camps. At one end were public cloud data centers offering a dozen or more site options in a regional context. At the other end were edge-focused providers, which typically emerged from content delivery network (CDN) companies and invested in thousands of small points of presence. We expect to see these models converge, with the focus on applications, use cases, and capabilities rather than infrastructure paradigms.
Serverless development platforms are at the heart of cloud-native technologies
Given the above trends, here is the new market definition: Serverless platforms are:
“A cloud-native software development platform that abstracts the underlying cloud infrastructure, complex server configurations, runtime characteristics, and deployment patterns from the development process. FaaS is the most common implementation of serverless development and forms the core of serverless architectures, but any platform that meets the definition is eligible. Serverless development platforms support the deployment of any business logic, decouple state from the underlying compute, automatically scale on-demand, offer micro-billing (often in milliseconds), run on managed cloud infrastructure, and support event-driven communication. Additionally, there are extensions that expand the use cases that serverless development platforms can address, such as state/storage services, distributed management infrastructure, asynchronous messaging, observability, and security.”
The original article is here.
The views and opinions expressed in this article are those of the author and do not necessarily reflect those of CDOTrends. Image credit: iStockphoto/VectorHot
