New providers specialized in front-end
The major cloud providers are not the only players in the serverless world. As early as 2015, just one year after the launch of AWS Lambda, Netlify released its own hosting service for sites built with the 'Jamstack' architecture (a web development pattern based on pre-rendering the entire front-end and exposing databases and custom logic through APIs).
Today, platforms of this kind combine CDN services (servers distributed in multiple geographic areas that speed up the delivery of web content by bringing it closer to users) and serverless with highly advanced development tools and the native ability to run applications or parts of them on the edge (the portion of the network closest to the user). They are integrated with major front-end development frameworks with server-side rendering (e.g., Next, Nuxt, Remix) and static page generation (e.g., Gatsby, Hugo, Astro), providing full-stack support for projects of this kind. The main providers are: Cloudflare, Netlify and Vercel. Among the characteristics that make these platforms ideal for web applications are::
- Automatic infrastructure configuration from the application repository;
- Automatic and multi-environment management of released versions, with rollback capabilities;
- Built-in services activatable on the edge, such as: firewall, routing, middleware functions, A/B testing, authentication
Our experience
Despite the widespread adoption and robustness of the serverless paradigm in enterprise contexts, adoption levels in the banking and financial services sector remain low. This is not surprising given that the migration of systems to major cloud providers is still in its early stages in this industry.
Among the projects already carried out in this area, we have observed the maturity of the services offered by Cloud providers with satisfaction from both customers and development teams. Among the most recent experiences, we implemented an AI system for optimization problem solving using Azure Container Apps services for all calculations. The scalability model and service configurability proved to be excellent and suitable for the use case. The cost model also proved optimal for the specific use case (where some jobs execute intensive computations for a few hours a day), compared to equivalent solutions based on PaaS or IaaS.
In general, we have had the opportunity to use all the major providers, noting significant differences and specializations in services. While AWS tools seem to be favored in the FaaS domain, Google Cloud and Azure offer excellent solutions in container-based solutions.