Introduction:
While managing a complex hybrid technology landscape, a leading BFSI company in India was facing significant challenges in controlling their cloud costs. With over 50 AWS services and various applications with diverse requirements, the company wanted to optimise costs, reduce operational expenses, and enhance margins. Manual operations, reliance on licensed tools, and a lack of automation added significantly to the issue, and it became imperative for them to efficiently manage their cloud spend while scaling up their business operations.
Business Challenge
The company's hybrid technology landscape comprised multiple applications with different technologies, operating systems, and availability requirements. With multiple lower environments shared across teams and manual operations, cost optimisation became increasingly complex. The lack of automation further made it difficult to control cloud costs effectively. Additionally, the company was using a lot of licensed tools that added to their operational expenses. This made it challenging to improve margins and enhance business growth.
Solution:
A Site Reliability Engineering (SRE) team with expertise in AWS, DevSecOps, security, and infrastructure as code was formed. The team adopted a structured approach, categorising the landscape and analysing each category using standard checklists. The cost optimisation project was divided into multiple categories, like pricing models, auto-scaling policies, automations, migration to cloud-native solutions, governance, scale-up policies, and monitoring.
Pricing Models:
Workloads were grouped into categories as lower and higher environments based on usage patterns and requirements, enabling the implementation of appropriate pricing models such as on-demand, reserved instances, and spot instances.
Invasive and non-invasive changes: Changes which do not need major changes in code or architecture and the ones which need are categorized.
Auto-Scaling Policies:
Auto-scaling policies based on metrics like CPU, memory, RPM, and peak traffic patterns were defined to ensure efficient resource utilisation and cost savings.
Automations:
Few categories were created for automations like repetitive manual tasks, DevOps tasks, cost optimisation tasks, governance and reporting tasks etc. Various technologies automations stacks were built based on this. For e.g.: Self-service portal for starting or stopping lower environments, DIY portal for tasks like monitoring health of ECS services or servers, start or stop of application servers with given count, governance reports like EC2 inventory, housekeeping tasks like auto-deletion of snapshots or AMIs, utilities to generate heap dump, and various Jenkins based automations were made.
Tools Development:
Tools for patching frameworks, AWS Spot frameworks, auto-healing of incidents, and process frameworks for monitoring cost and infrastructure capacity were developed to automate tasks and improve efficiency.
Impact:
The implementation of the comprehensive solution resulted in a significant reduction in average monthly cloud costs, with savings ranging from 25% to 50%. The company gained visibility and predictability in cost forecasting, tracking, and optimisation, allowing them to control cloud spend effectively while scaling up their business operations. By adopting a holistic approach to cost optimisation and automation, the company was able to enhance margins and achieve greater operational efficiency in managing their hybrid technology landscape.
