The (continued) rise of Platform-as-a-Service
One of the biggest areas of growth over the past year was not just overall cloud adoption, but the transformation of traditional virtual machine usage in the cloud to serverless or Platform-as-a-Service (PaaS).
Even small-to-medium businesses adopting public cloud include at least 30% serverless on average in their cloud ecosystem. This stat is important as it highlights that this is not just an enterprise-level solution or development and has a much wider usage spectrum.
This is with good reason, as businesses begin to modernise, seeking greater returns - and with the rise of remote working a fear of being left behind in many industry verticals is a growing concern.
So what’s driving this move away from a traditional virtual machine-based architecture?
Scaling
Fluctuation in demand has always been something businesses have had to adapt to. Traditional architectures suffer from finite resources and challenges adding capacity on demand; cloud serverless architectures scale much faster and simpler than before. With serverless nearly all components can have capacity increased or decreased with very little notice, often no downtime and with much less reliance on architectural changes or risk.
The risk associated with scaling has often been a misunderstood or under prioritised area in the past. Increasing or decreasing VM or even physical server-based architectural components could cause a system problem or failure which could have significant impact on the business. With serverless all these issue are non-existent to the consumer as serverless is designed with this in mind and is handled automatically on the customer’s behalf by the cloud provider.
The entire ecosystem is designed with on-demand scaling, high availability and redundancy at its core and are handled much more seamlessly than is often possible from your own traditional architecture, capability and budget.
PaaS/Serverless continues to see a considerable growth rate with compound annual growth rate (CAGR) of 21.7% from 2021 to 2028. By 2028 it’s expected to hit a revenue of $36.84 billion worldwide, from $7 billion in 2020.
Verified Market Research® Serverless Architecture Market 2022 Report
Development speed & agility
One thing that is evident over the past 5 years is the ever increasing need to get product features or systems changes released as quickly as possible to gain competitive advantage or just to keep up with the competition. This is where serverless and DevOps methodology come into their own.
Serverless requires DevOps to accomplish this with code releases being applied to the serverless components - these two areas are designed to go hand-in-hand. With serverless, every part of the architectural process is defined by its own components, introducing a modularised approach which avoids wide ranging impact of changes and therefore reduces testing requirements.
This approach sees businesses move away from periodic centralised releases to adaptable release management – resulting in shorter development times and increased deployment frequency, whether that's taking products to market or improving internal systems.
Reduced administration
While best practice dictates that hardware should be replaced every five years as mean-time-to-failure reaches a critical point, most hardware is pushed well beyond that as businesses look to gain maximum return out of their capital expenditures. The issue here is on-premises hardware performance and quality erode on average 14% every year. By Year 5 a typical hardware set up may only be offering around 40% of its original capability.
Migrating to a cloud service with virtual machines was the next logical step and this alleviates some of these issues, but you’re still left with the administrative burden around uptime, patching, architectural and multi-dependency complexities.
Serverless technologies reduce this administrative overhead and much of the burden is eliminated. With no virtual machines to patch, traditional VM security is no longer a concern, even capacity considerations to a degree are significantly reduced making serverless architectures increasingly popular.
Your dependencies also become compartmentalised making it much easier to focus on change and improvement without duly impacting on other services using the same underlying capacity.
On-premises hardware performance and quality erode on average 14% every year. By Year 5 a typical hardware set up may only be offering around 40% of its original capability."
Cost
With platform and serverless resources, cost is much more tightly controlled and often much cheaper than running traditional virtual machines.
This is due to serverless architecture components each being costed and available for a very specific requirement and, by its very nature, already right-sized. By changing your architecture to utilise on-demand components, the meter is running only when those components are required.
Serverless in action
To demonstrate, let’s take a traditional application that receives an instruction and performs an action on its internal database. The application needs to be highly available, online 24 hours a day, 7 days a week (just in case), and on average is busy, preforming around 300 actions per hour, for the first and last 30 minutes of the day.
By transforming this to an on-demand serverless service, the cost is dramatically reduced and becomes variable as shown in the diagram below.
By providing flexibility and multiple routes for customers to explore, it has never been easier to get going on this part of a digital transformation journey.
