Amazon Web Services (AWS) believes IT infrastructure management is evolving too rapidly for the average IT organization to keep pace.
Starting with the deployment of containers in a production environment, Werner Vogels, AWS CTO, contended at a recent AWS Summit event in New York that because of the extensive amount of existing tooling AWS has put in place to manage workloads, it makes a lot more sense to deploy containers on top of virtual machines.
To prove that point, Vogels noted that both a new advanced load balancing service for AWS and a new Amazon Kinesis Analytics service for launching SQL queries against data in AWS can be used with applications running directly on virtual machines or deployed on containers.
Besides eliminating the need for new tools to manage containers, Vogels suggested there was much to be gained by making it simpler to integrate container applications with legacy applications already running on the AWS platform.
But Vogels didn’t stop there. AWS has been investing in a serverless approach to running cloud applications using an event-driven architecture known as AWS Lambda. Vogels suggested that, in time, many of the developers opting for dedicated container-as-a-service platforms today would opt for what Vogels described as a “functions” approach to delivering microservices, which would eliminate the need to provision and manage servers altogether.
In fact, Vogels noted that AWS would be the only public cloud platform capable of providing a common base of management tools across traditional virtual machines, containers running on virtual machines and a serverless computing platform. From a DevOps perspective, Vogels said the serverless computing approach replaces a “pets versus cattle” approach to managing servers with a “herd” mentality, where the existence of a server is only relevant within the context of a massive pool of always available IT infrastructure resources.
A serverless platform can be applied to containers, virtual machines or any other atomic unit of computing that emerges in the future. Vogels isn’t saying containers will go away as much as he is making a case for their underlying IT infrastructure becoming invisible. But he is suggesting that much of the fierce DevOps debate over how best to manage those resources will become moot.
At the same time, Vogels is also making it clear that, going forward, internal IT organizations will have a difficult time keeping pace with the rate of IT infrastructure management investments required to make a serverless computing service available to developers. As such, Vogels is betting that even more application workloads will be moving into the cloud.
Of course, AWS is not the only cloud service pursuing serverless computing architectures. Google has created its Google Cloud Functions service, while Microsoft is currently providing limited access to its Microsoft Azure Functions service. None of this means that serverless computing frameworks won’t one day come to the enterprise. It simply means that the rate at which new computing architectures can be deployed on a public cloud for the moment far exceeds anything internal IT organizations can ever hope to match.