The serverless architecture enables unprecedented scalability and agility for businesses looking to outperform in a highly digitized environment. As businesses increasingly adopt microservices or serverless computing, its strengths and weaknesses are becoming clear. However, some questions remain: will businesses need Infrastructure Services at a time when the infrastructure itself is becoming invisible? Will they need the services even if all legacy architecture is modernized using cloud native architecture? The answer to both questions is ‘yes.’
Let’s explore four key characteristics of serverless computing that necessitate the deployment of new Infrastructure Services to ensure success:
#1 Increases operational complexity compared to monoliths or n-tier
While serverless architecture enables greater flexibility to developers in terms of change and release management, it also brings greater operational challenges. Calling multiple remote services to execute a given request is fraught with complexity - should you call them in parallel or serially? The data pathways are also significantly complicated with innumerable data exchanges and crossovers. What about all the errors (both at the application and network level) that could arise at any point in the chain, and what would be their impact on the request? It’s clear that distributed transactions require a different approach to provisioning and handling of errors.
#2 Drives innovation but complicates the journey from innovation to stability
For enterprises that have a fail-fast mentality, microservices is the ideal approach as it helps enterprises iterate, fail, and recover quickly. But as the enterprise moves forward to the maturity and stability stages, a business’ appetite for risk typically decreases as the cost and scale involved are significantly higher. When attempting to scale microservices, tooling, troubleshooting, data persistence, and consistency can become highly convoluted.
#3 Makes business process integration tedious and time consuming
For fully cloud-based enterprises, integration isn’t much of a challenge. But for those comprising a variety of architectures - from mainframes to legacy Unix/Windows/Linux systems - it is a tall task. The tedious and time taking procedure of refactoring applications into cloud native formats introduces a new API service exposure layer during integration, further aggravating complexity. The integration problem isn’t just restricted to applications – it extends to the management and operations toolset as well. This raises the question: how to integrate identity and change management, incident integration, and backup, restore, security systems, etc. across different architectures?
#4 TCO calculation requires a radically different approach
The way enterprises usually calculate TCO for server-based applications is - identify the cost elements, annualize the CapEx, add OpEx, and finally arrive at an annual TCO. In microservices or serverless architectures, it is not the investments but the usage patterns that constitute the cost elements. Additionally, because usage consumption varies across timeframes, TCO will not stay the same across years - further muddling financial projections, modeling, and planning.
Key role of Infrastructure Services in driving serverless computing success
While cloud native computing helps businesses drive unbeatable agility and scalability in a demanding digital world, it’s clear that it can bring significant challenges and complications in its wake. Deploying relevant Infrastructure Services - manual as well as bot-driven - that are geared to the cloud native environment can spell the difference between success and failure. Tightly integrating the Infrastructure Services with financial governance is also paramount to driving revenue linked cost management and financial accountability across the organization, enabling superior outcomes and profitability.