Overview
By migrating PLM workloads to AWS, manufacturers can improve time to market by increasing the performance of PLM workflows across global remote engineering teams. A motor vehicle manufacturer achieved this by partnering with Birlasoft.
Challenges
- There was a need to move the application away from legacy hardware. This involves upgrading the application to its latest version available in the market, decommissioning the current older version and deploying the application in a virtual environment where scale-in and scale-out are easy.
- It was necessary to ensure scalability, reliability and global reach for the customer’s end users. PLM migration to AWS was intended to achieve these.
Solution
The project involved provisioning for required infrastructure on AWS for PLM Windchill application, 5 TB of Oracle data migration to AWS RDS, 90 TB of application files migration to AWS S3, few GB of PostGreSQL data migration to AWS RDS and upgrade of application on Cloud to its latest version.
AWS Direct Connect from on-premises to Cloud was used to transmit data from on-premises to AWS. Infrastructure was provisioned through Infra-As-Code approach using Terraform. Custom scripts were used to automate the migration of application files to AWS S3. This migration also involved upgrading of Oracle database from 11i to 19c. Birlasoft automated end-to-end provisioning of infrastructure on AWS through terraform and orchestrated the deployments through Github Actions. Birlasoft also developed Python code for serverless service that generates patch compliance reports. CloudTrail logs and VPC flow logs were integrated with McAfee SIEM for security monitoring. All redhat servers were protected through intrusion detection pursing redhat's Advanced Intrusion Detection Environment (AIDE). Application upgrade activity happened in 3 iterations to ensure all the application features are customized on its latest version.
Development, Upgrade Test, Test, UAT, Pre-prod, Prod, DR environments were created on AWS.
Backup:
AWS Backup service is used to backup the Application and Database servers, with below policies-
- Daily backup with retention 15 Days
- Weekly backup with retention 12 Weeks
Restore:
The restore options used will replace the existing instance with new instance from automated backup-
- Point in time restore
- Restore from daily backup
- Restore from weekly backup
AWS Services used - AWS S3, EC2, AWS WAF, AWS VPC, Application load balancer
The Impact
This application is rich in features for manufacturing component design. The application processes millions of design files. The files are stored in S3 buckets. The main service level objective is that business should be able to access the designs and process them without any latency. Oracle database hosted on RDS would have pointers to the design files and application would present the content from files identifying them through these links.
- The tuning was done at various levels to achieve the required performance
- Oracle database was optimized, and the tables were built with relevant indexes for faster query performance
- Private links were created between S3 buckets and VPC to minimize the file access latency
- All the components in each of these layers were built in HA mode across two availability zones
- All the application servers were launched on EC2 instances and Database is deployed on RDS-Oracle platform
- Cloudwatch custom and default metric were used for various workload telemetry
- The overall application infrastructure design falls under three layers - App layer, Backend layer, Database Layer
- For health measures cloud watch metric were configured
Other Case Studies
