Going Server less for Web Services
summary
A genome-search web service shards whole-genome off-target searches into hundreds of AWS Lambda workers behind an API Gateway and DynamoDB, cutting infrastructure cost versus always-on beefy instances; GT-Scan2 is production-used by two Australian research centres. AWS X-Ray reveals a slow on-target predictor that is rewritten in Python and machine learning to drop runtime ~80% (two minutes to ~30 seconds). A separate CryptoBreeder pipeline auto-retrains on EC2, stores models on S3 and serves predictions via Lambda for under $25 in two months.
AI-written summary · about
chapters (11)
- 0:00Start
- 4:21Ideal application case for serverless
- 12:33Interoperable Workflows
- 13:48Serverless systems are hard to optimize
- 14:40GT-Scan2 X-Ray Analysis
- 16:25Results - 4x Faster (80% improvement)
- 20:07CryptoKitties in a nutshell
- 27:51Clinical use of GenPhen DB
- 28:54PhenGen Database
- 30:09Three things to remember
- 31:23Transformational Bioinformatics
related talks
42mAWS Lambda from the trenches
47mAdvanced Serverless Architectural Patterns on AWS
43mA Year of Diagnosing Pediatric Cancer with Event-driven Architecture
41mUsing Serverless & ARM64 for Real-Time Observability
38mBuilding Practical, Cost-Efficient GenAI Solutions Using Serverless
01h 00mServerless - reality or BS - notes from the trenches
44mServerless – Automated Tests of web apps in the Cloud world
47mServerless in production, an experience report