Initial scenario: E-commerce server with ~20 million transaction per day. Frequent issues of system slowing down, unresponsive at times and poor performance resulting in loss of business. Unable to identify and filter real transactions and quarantine fraudulent transactions and data. Batch jobs running for 15 minutes every hour and other transaction as well on the system. Spent of about $7000/per month on AWS bill.
In order to sustain performance and reliability, infrastructure was being scaled up further, increasing AWS costs every month. However business was not scaling up, partners and customer were unhappy, application was crashing very often and business was significantly impacted.
Stanra Advisory Services: Stanra consultants did a comprehensive analysis of the application database and AWS infrastructure with respect to the transaction volumes, historical data size. Consultants team carried out the following transformations in a phased manner
- Comprehensive review of the database, queries identifying the bottlenecks, fine tuning of many queries, adding appropriate indexes, date wise partitioning of the transaction tables, moving of older partitions to historical table through partition swapping.
- Elimination of batch jobs through Implementation of alternative continuous processors.
- To rationalize the AWS infrastructure among many other activities, following major activities were carried out
- Eliminated provisioned IOPS from 25000 to zero
- Scaledowned the two EC2 and one RDS server from 4x to 2x
- Implemented Autoscaling
Current state: 100% application availability, System performance at 5x and a cost at 25%.. Full automation of housekeeping jobs with no manual intervention whatsoever and routine maintenance/support time reduced to 25% through automation.
Need help on cloud based solutions, drop a mail at firstname.lastname@example.org with your requirements
Leave A Comment