A Guide to Serverless Architecture

The tech industry thrives on the new. Concepts are driven, innovations are pushed, but sometimes execution can fall somewhat short of lofty blue sky – or in this case perhaps blue cloud – thinking. However, serverless computing is projected to undergo some phenomenal growth.

Serverless architecture has been in mainstream computing consciousness for just shy of a decade; Amazon Web Services launched their platform Lambda in 2014. At the end of last year, the market was worth around US$9bn, boosted by new providers such as Google Cloud Functions, Vercel, and Cloudflare Workers.

Forecasting gurus Global Market Insights are predicting that in the forthcoming decade, that figure will increase by a factor of ten to approximately US$90bn by the close of 2032. But what are the tech pros that are making serverless architecture so attractive to the tech pros? Let’s have a look.

Apps and Applications

Serverless architecture works on the principle of Function as a Service (FaaS). Use is provided on demand, rather than having a server continually whirring. It expands past cloud computing’s more traditional Product as a Service (PaaS) model by being event-driven and almost infinitely scalable; a necessity when we countenance the increasing impact of edge computing and the Internet of Things into every facet of modern life.

Serverless architecture is increasingly being utilized by blue chips, with Netflix, Nordstrom, and T-Mobile among companies taking a serverless first approach. Coca-Cola has employed serverless technology to run payments on their vending machines, bringing the cost of annual running on each machine from US$13k pre-adaptation to US$4.5k now.

Serverless providers can offer big benefits to smaller outfits also. Some will provide Backend as a Service (BaaS), meaning that a fledgling developer can focus on coders working on applications rather than spending time on database management.

As MongoDB’s post on serverless architecture explains, serverless frameworks also put pay to making a decision about cloud-hosted database sizes from a set of pre-determined packages. The company simply starts a database and it can grow as necessary as data and throughput expand.

Up(sides) in the Cloud

First, serverless architecture isn’t actually serverless. They’re still needed by the business that’s using the infrastructure. However, those servers are located in a data center owned by their cloud computing provider. This has obvious upsides for the startup.

The hefty initial outlay is negated, as is space to store a physical server. Upgrading as the company expands also becomes a non-issue with the non-presence of servers, saving time (and money) on DevOps and capacity planning. Serverless providers work on a ‘pay as you go’ model, charging only for server usage as it happens, with some so precise they bill in 100 millisecond increments.

The aforementioned Coca-Cola vending machines going serverless are estimated to save the company around US$25m annually. Code can be uploaded in increments or all at once, making it quick and easy to release, update, or patch software. Swift patching and updating will certainly save cost and time in customer services!

Is Serverless for Everyone?

However, as well as uploads, there are downsides. Debugging and testing can be complicated by the end user not having access to backend ops and processes. The company using serverless architecture doesn’t control the security of their storage, which may be an issue for those handling sensitive data, however, concerns have been somewhat assuaged by the 2019 launch of the Confidential Computing Consortium, including key players like Google, IBM, and Microsoft.

The increasing public awareness and adoption of 2FA have also helped curtail enormous, resource-consuming password databases. Provider ‘lock-in’ can be an issue for smaller companies as moving from, say, AWS Lambda to Google Cloud Functions can be strenuous.

Some platforms, however, such as Atlas, let you code into the framework of your choice, meaning it’s less of an issue.

Serverless architecture is certainly here to stay. The behemoths involved in the provision of the function mean that almost infinite resources can be dedicated to the applications of the future. However, in doing so, they’re enabling startup developers to get their ideas out there more quickly and with less onerous up-front costs.