Data Driven Scalability and Cost Analysis for Evolutionary Architecture In the Cloud
summary
Runs systematic cloud experiments on Google App Engine using a Java load client and 1,024 concurrent clients, comparing Go, Node, Java and Python under a write-heavy profile and costing over $10,000 to execute. Finds Go delivers the best latency and about four times the price‑performance of Python (Go ~17 instance‑hours vs Python >50). Shows App Engine defaults (CPU 60%, max concurrent 10) are suboptimal and tuning CPU and max concurrent cuts costs by 20–46% while keeping ~96–99% throughput. Publishes raw data for reuse.
AI-written summary · about
chapters (9)
- 0:00Start
- 2:45Market Research about Cloud Adoption
- 10:20Database Storage Tier Costs
- 10:58Database Consistency in Dynamodb
- 12:58Google Cloud Platform
- 19:46Calculate a Price Performance Ratio
- 28:56Performance
- 31:36Swagger
- 32:27Summary
related talks
01h 25mWhat is Google App Engine?
01h 22mWhat is Google App Engine?
20mAtomics Habits in cost optimizations in the cloud
20mCollision 2015 - Urs Holzle of Google with Bruce Upbin of Forbes
01h 03mLaunch and grow your startup on GCP
36mMigrating Applications to the Cloud in a Cost Effective Manner
57mDebugging Java in Production on Google Cloud Platform
36mGoogle Cloud Platform: Cloud Functions, App Engine vs Container Engine.