What Is Serverless Computing?

Serverless computing is a type of cloud computing model. In cloud computing, a cloud services provider manages computing infrastructure like servers, storage, and networking, providing access to organizations and users via the internet. The cloud provider is responsible for maintaining, updating, and securing the service, and the customer can consume cloud resources on an on-demand, pay-as-you-use basis. In serverless computing, the cloud provider is responsible both for the cloud infrastructure as well as the scaling of apps. Serverless applications are typically deployed in containers that can run in any computing environment and can launch on demand when called.

With serverless computing, developers can focus all their energy on writing front-end application code and business logic, deploying it to containers managed by the cloud service provider. The cloud provider spins up and provisions the required resources on demand when an application’s code is executed, scaling back to zero when execution stops. When an event triggers a serverless app to run, the cloud provider dynamically allocates and scales resources for the code, ending allocation of resources as soon as the app stops executing.

Function as a service (FaaS) is a cloud computing service that lets developers run code or containers without managing the infrastructure required to run the code. FaaS is an essential part of the serverless computing model, but serverless computing offers much more, including an entire stack of services such as serverless databases and storage, event streaming and messaging, and API gateways.

• Reduced costs. Serverless computing is available on a pay-per-use model, where development teams can pay only for the services they use. This can be quite cost-effective compared to traditional methods of provisioning infrastructure for development environments, which can involve a great deal of over-provisioning and underutilization of assets.
• Immense scalability. Serverless technologies can scale easily to accommodate demand and business requirements, while developers don’t have to worry at all about managing policies needed to scale up the code.
• Accelerated time to market. Serverless architecture can significantly accelerate speed to market by enabling development environments to be provisioned incredibly quickly.
• Greater productivity. Serverless solutions let developers focus on writing code, rather than managing infrastructure, giving them more time to innovate their front-end functionality and business logic.
• Increased flexibility. Serverless platforms allow developers to write code in any language or framework.
• Faster deployment. Developers can access serverless computing resources in minutes or hours, rather than waiting days, weeks, or months for on-premises infrastructure to be installed.

• Potential latency. When a scalable serverless platform handles a function for the first time, there can be a delay or a “cold start” that may cause latency.
• Difficulty debugging. Serverless instances make it more difficult to collect the data needed to debug and fix a serverless function, since serverless instances create new versions every time they are spun up.
• Higher costs for long-running apps. Long-running tasks may cost more on a serverless environment than on a dedicated server or virtual machine.
• Vendor lock-in. Relinquishing control of the underlying infrastructure of an IT stack can open up an organization to potential vendor lock-in, where changing providers may be prohibitively expensive.

Like serverless computing, platform as a service (PaaS) offerings allow developers to access application development tools without worrying about managing the back end. However, PaaS solutions do not scale as easily as serverless computing solutions, and the pricing model of a PaaS offering is not as precise as serverless.

Kubernetes is an open source container orchestration solution for managing and launching containerized applications. In a managed Kubernetes solution, third-party providers are responsible for some or all of the tasks involved in setting up and operating a Kubernetes platform. While serverless solutions offer little access to the underlying infrastructure, a Kubernetes solution provides full control over the application hosting environment. Serverless solutions scale more easily than a managed Kubernetes environment, and a pay-per-use pricing model may offer greater cost savings than Kubernetes.

Back-end as a service (BaaS) solutions are similar to serverless in that they allow developers to rent back-end services that allow them to avoid managing servers and back-end infrastructure. However, BaaS solutions may not automatically scale an application, and because they are not event-driven like serverless solutions, BaaS offerings may require more server resources.