AWS Cost Optimization with Graviton Migration

Leveraged AWS for seamless Cloud Migration, DevOps, and Optimization

A detailed AWS Cloud optimization and cost-saving initiative was executed for the application's Production and UAT AWS environments, focusing on migrating compute to the Graviton (ARM) architecture, implementing data lifecycle management, and enhancing security measures for the database infrastructure.

Customer Overview

Consumer Services Customer

The customer is a rapidly scaling consumer experience platform that delivers high-volume, short-form content, real-time updates, and mobile-first entertainment to users across India. The customer required a strengthened cloud foundation with improved performance, cost governance, and security to support sustained growth and operational reliability.

Location

Bangalore Urban, India

Company size

2-10 Employees

Problem Statements

We identified key problems that need to be addressed to deliver a solution aligned with the customer’s problem

High Operational Costs

Running compute-intensive workloads on x86 (AMD) architecture instances and over-provisioned EC2/RDS instances resulted in higher-than-necessary monthly cloud spending.

Unoptimized CloudWatch Log Storage

The majority of CloudWatch log groups in the Production account were set to "Never expire," leading to growing and unmanaged log storage costs.

Inefficient Data Storage Management

Lack of lifecycle policies for ECR images and S3 buckets resulted in unnecessary storage costs from older image versions and un-tiered object storage.

Application Downtime

Management Critical implementation steps, such as EKS node group changes, RDS downsizing, and EC2 downgrading, required careful planning and execution to minimize application downtime.

Our Methodology

An insight-driven approach focusing on diagnosing usability gaps, redefining user flows, optimizing visual hierarchy, and executing scalable design solutions to enhance functionality, engagement, and decision-making efficiency across digital touchpoints.

01 Discovery & Diagnosis

Reviewed workloads in the Prod and UAT AWS accounts for cost, performance, reliability, and security.
Identified that EKS clusters in both environments were running on x86 (AMD) architecture.
Confirmed that 33 out of 36 CloudWatch log groups in the Production account had a “Never expire” retention period.
Identified ECR, S3, and old support plan as areas for cost optimization.

02 Solutioning

A comprehensive strategy was implemented across the application's infrastructure, covering EKS migration to ARM, RDS modifications, EC2 downsizing, CloudWatch log retention changes, ECR lifecycle policies, S3 lifecycle policies, and AWS Support Plan downgrade.

03 Execution & Delivery

Production EKS nodegroup migrated from AMD to m6g.large; UAT EKS workload also migrated to ARM with buildspec.yaml and CodeBuild configurations updated.
RDS Dev instance downgraded, and storage migrated from gp2 to gp3; public access disabled and subnet groups moved to private subnets.
RDS Prod instance changed to db.t4g.large and storage upgraded to gp3.
PostgreSQL upgraded from 14.13 to 16.4 for dev and prod.
EC2 instances right-sized including the App-DOCKER-PROD and Prod-EKS-Management-Server.
CloudWatch log retention updated (Prod: 12 months, UAT: 90 days).
ECR lifecycle policy implemented; 115 older images removed.
S3 lifecycle policies added to transition data to Intelligent-Tiering and expire log bucket objects after 60 days.
AWS Support Plan downgraded from Developer to Basic.

04 Post Implementation

Business Impact

We delivered measurable business outcomes that validate the effectiveness of the solution

48.75%+

Reduction in Overall AWS Cloud Spend

92.60%+

Improvement in Security and Compliance Posture

44.25%+

Optimization in Storage and Log Management Costs

33.85%+

Improvement in Database Performance and Reliability

Relevant Industries

Consumer Services

Problem Statements

Business Impact