Rajat SharmaSVP and Global Head of Technology Ecosystem and Growth for Zensar Technologies.
Let’s discuss the most pressing need for businesses in today’s business environment, where the cloud is being used extensively as an engine of agility, innovation and growth: Cloud costs are becoming increasingly unmanageable.
When looking at cloud costs, they are never fixed and vary as each business is at a different stage of their digital transformation journey. They are either using the cloud for burst capacity, moving more workloads to the cloud or enabling newer features like serverless, IoT or AI/ML/GenAI in the cloud to become cloud native.
Similar to how the cloud enables the ability to synthesize IT (infrastructure, data, security, and applications) through a programmable core, costs must also be dynamically measured and controlled with multiple moving parts through three key pillars—governance, automation and architecture. This is what I call “cost composability”.
Pillar 1: Governance
Most organizations focus on FinOps with a single “governance” lens, starting with how cloud services are procured—types and nature of instances. how do you consume them? the number of accounts, functions and environments; how they are tagged and monitored; and finally, how different cloud providers charge you.
They then establish policies, governance and guardrails around these aspects by purchasing a FinOps/CMP tool to monitor usage and automate policies. This is necessary and has become a prerequisite for today’s CloudOps.
Pillar 2: Automation
Now, let’s look at the key driver of cost reduction in hybrid operations, which is relentless “automation”.
This covers a wide range of fundamental IT tasks, including disk cleanup, password reset and patching, alongside comprehensive workflow and workload management. It extends from IT service management to batch automation, culminating in the adoption of true DevOps principles. This holistic approach facilitates continuous improvement, ranging from infrastructure-as-code to complete full-stack observability supported by ML and GenAI models.
This area has matured and seen continuous improvement in recent years, with IT outsourcing and technology partners helping businesses move to predictive functions through relentless automation.
Pillar 3: Architecture
The third pillar, architecture, includes future-proofing design and technology. Many customers are unaware that architecture plays a critical role in managing, controlling and optimizing cloud spend. Most cloud providers recommend and support the implementation and adoption of WAF (well-architected framework) for defining and planning cloud workloads.
In public cloud environments like AWS, Azure, and GCP, dozens and hundreds of features and functions are released every year. Therefore, we cannot afford to have a fixed target architecture and operating model. Instead, it requires an ever-evolving future architecture and operating model that needs continuous improvement.
There are four key areas of architecture and operations that must be addressed to achieve cost composability (optimizing costs and maximizing cost of uptake).
1. Capacity and performance: How do you continue to measure and manage the capacity and performance of the resources your applications are using? This includes monitoring I/O operations between memory (storage) and CPU, as well as read/write ratios. In addition, it involves monitoring the percentage usage of resources such as CPU, memory and storage through services such as EC2, S3 and EKS clusters for both peak and average load.
2. Number of workload environments: What is the typical ratio of non-production to production costs? How can you continue to move these ratios towards the magic number “1” where production and non-production are in 50:50 balance? We’ve noticed that most clients operate with a 70:30 ratio, resulting in a ratio of 2.3. This can be achieved by reducing the number of low-level environments for development and maximizing automation in testing and staging processes
3. Workload and Layer Types: We’ve seen the majority of customers run 70% to 80% of their workloads using only Infrastructure-as-a-Service (IaaS) functions, mostly in VMs. This is mainly due to the initial wave of lift-and-shift migrations, which resulted in significant consumption of resources and consequently led to exponential increases in costs.
Each workload/application must be thoroughly evaluated for a strategic shift from IaaS to PaaS (Platform-as-a-Service) or FaaS (Function-as-a-Service)/serverless and containerized. Additionally, static content such as websites can be shifted to run in S3 buckets instead of using EC2 instances, and the same can be applied for real-time storage archive access or large storage in Redshift.
4. Protecting the future of technology: Here, the business and service provider partner work together to review new features and utilities from the cloud and third parties to adopt new technologies while optimizing costs and creating a competitive advantage over industry peers.
A typical example in today’s world would be leveraging a next-generation Microsoft stack—Azure Synapse, AzureML, Copilot, and OpenAI’s GPT—or a next-generation AWS Stack—Glue, SageMaker, Bedrock, CodeWhisperer, and Q—to build and run LLM rather than trying to build AI/ML models on the core stack (IaaS) and GPX with unoptimized third-party LLMs.
Achieving Cloud Cost Composability
The three key elements—governance and compliance, automation, and architecture—must be coordinated to effectively execute and optimize the entire cloud lifecycle. This includes everything from strategy and assessment to cloud operations, including foundation and migration, covering DevOps, SecOps and FinOps.
FinOps should not be an afterthought or a standalone activity done ad hoc. It is a continuous operation that requires all three sub-functions—governance, automation, and architecture—to participate in the FinOps office to achieve “cloud cost composition.”
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Am I eligible?
