It’s fair to say that the use of containers to manage cloud infrastructure is now firmly mainstream. But just as it reaches maturity, we may have already seen its usage peak. This may seem a counterintuitive thing for me to say, especially as containers are one of the main things keeping us busy right now, but what’s clear is that their days are undoubtedly numbered.
The reason being that serverless, an event-driven alternative, is providing businesses with a far superior option – one that will leave containers redundant in the long term. Moving to a transactional model can not only considerably reduce cloud costs, but it enables costs to scale linearly with usage. The savings this offers simply cannot be ignored.
In addition, adoption means developers move even further up the shared responsibility model. They no longer need to worry about infrastructure updates, scaling or security – they can, therefore, shift resources to the development of new features. This will allow them to focus on improving the business, rather than just keeping the lights on.
The benefits are so abundant that if you have developers not yet writing for serverless, you need to hit the pause button and ask some questions.
You might find challenges in adoption at first, but it’s only a matter of time before this becomes the de facto option for all businesses – and the need to containerise, as a result, will begin to fade away.
This is a seminal shift but then that’s nothing new when it comes to infrastructure – it’s not that long ago that people stopped writing for on premise and embraced cloud. Yet now, you wouldn’t have developers writing for anything but a public cloud environment – it’s the default place to run a business.
The serverless scenario, however, is probably more akin to one we experienced a couple of years ago when containers started to enter the mainstream. It meant that we could all adopt a standardised approach. So, instead of tooling our own packaging and deployment workflows from scratch every time, we had an answer straight out of the box. In the same way, projects such as SAM, Serverless and more recently CDK are already providing superior frameworks to start building and deploying cloud native applications.
With any major industry movement of this nature though, the change isn’t going to happen overnight. When you still have existing apps, the reliance on containerisation will continue for several years to come. A move away would involve rearchitecting from the ground up, and that’s simply not an option for most businesses. Instead, start small - introduce new serverless architectures to solve business challenges as they arise.
Unusually, we are likely to see the shift over to serverless being led by enterprise scale organisations – a rare occurrence when it comes to cloud innovation. But they are the organisations with the resources to do this, whilst not impacting ongoing project deadlines. Startups are another area in which we expect to see a ramp-up in Serverless adoption, with smaller development teams driving to bring their products to market quicker than their competitors.
That's not to say SMEs won’t be able to enjoy some of the benefits of serverless in the short term. What we’re likely to see is an interim period where organisations deploy services, such as AWS Fargate, which can bring some of the benefits of serverless, with little architectural changes required.
It means that any organisation that wants to shift resources away from managing infrastructure right now can do so, bypassing responsibility to Amazon to manage on their behalf and enjoy the worry-free capacity this provides.
With both ECS and EKS now supported on Fargate, there's little reason to deploy and manage this infrastructure yourself anymore.
The end of containers?
Whether organisations adopt Serverless or Fargate, what’s obvious is there is a clear desire for companies to shift their internal focus toward dev work. It makes sense if you want to concentrate more resources on creating innovative solutions that will give your business a competitive advantage.
The need to manage infrastructure will never disappear; there will always be significant responsibilities which need to be handled either internally or outsourced to a managed service provider. But as time goes on, we see the area of responsibility for cloud customers move further and further up the stack.
What’s certain though, is that the desire to move away from managing infrastructure is incentivising the shift to serverless – and containers are going to be swept aside in the process. With many devs already writing for this environment, and Fargate already being viewed as just a stepping-stone towards this eventuality, industry-wide adoption of serverless is inevitable.