Enable processing capacity and scale up with AWS Lambda
summary
PowerSchool migrates a paper-scan benchmark pipeline from datacenter servers and NAS to S3, Aurora, AWS Lambda and AWS Step Functions, replacing 300+ scan engines and FTP servers with managed Transfer Family and serverless functions. Monthly ops cost falls from about $15,000 to $200, end-to-end scan latency drops from ~30s to 12s at peak, and overall ops spending reduces by 97.6 percent, while parallel throughput scales into the thousands.
AI-written summary · about
chapters (19)
- 0:00Introduction
- 1:50About PowerSchool
- 2:57Student Assessment
- 5:53Architecture
- 7:42Scaling
- 9:14Deployment
- 9:28Migration Strategy
- 11:22Traditional Migration Approach
- 12:59Lift Shift Architecture
- 14:43Lift Shift Benefits
- 15:32Increased Scalability
- 15:43Software Maturity Model
- 16:41Why AWS Lambda
- 18:47Step Functions
- 19:18Database
- 19:43Assessment Scan Architecture
- 21:34Benefits
- 26:53Wrap up
- 27:34Session Survey
related talks
41mUnlocking Faster & Efficient Data Processing w/ Serverless
36mTransforming SaaS with Serverless Architectures
54mFeature management: De-risking migrations and releases for PowerSchool
32mServerless Data Processing and Machine Learning
47mWhat We Learned Going Serverless
01h 04mMigrating to Serverless - an experience report
46mHighly Scalable Image Storage with AWS Serverless
39mServerless in production, an experience report