Perfect Scalability
summary
Defines “perfect scalability” as near‑linear scaling: double nodes, double throughput, and shows where real systems hit a wall (single load‑balancer or shared DB). Identifies four enemies—shared state, ordering, chatty communication, failure—and prescribes concrete fixes: partition namespaces across regions, give services their own data, use event sourcing and command sourcing (persist incoming commands to a queue, ack immediately, process asynchronously), favor CRDTs/CQRS, avoid distributed transactions, and rely on Akka/Lightbend plus Docker/Ansible/Mesos/Kafka/Spark and deep monitoring for elastic, resilient scale.
AI-written summary · about
chapters (32)
- 0:00Introduction
- 1:34Defining Scalability
- 4:09Perfect Scalability
- 7:11Where would we fail
- 8:16How would we fix it
- 9:04What we did
- 9:20Dolls Law
- 10:36Universal Law of computational scalability
- 11:24The wall
- 13:08Avoiding the wall
- 13:32Limitation
- 15:05Avoiding Sharing
- 15:41State
- 16:51Communication
- 18:04Sequence
- 20:17Avoiding Sequence
- 21:26Design Changes
- 22:34Scalable Architecture
- 23:26Elasticity
- 26:31Spike Load
- 28:01Command Sourcing
- 29:13Graceful Degradation
- 30:15Microservices
- 33:31Persistence
- 35:27Distributed Transactions
- 35:56Idempotency
- 36:54DDD
- 37:54Other Concerns
- 38:23Verification
- 39:36DevOps Matters
- 41:01Conclusion
- 41:37Questions
related talks
32mAn Uncoordinated Approach To Scaling (Almost) Infinitely
43mThe Math of Scalability
44mDistributed systems and scalability rules
30mScalability Is Quantifiable: The Universal Scalability Law
22mWhy Are Distributed Systems so Hard?
29mScaleCheck: A Single-Machine Approach for Discovering Scalability Bugs...
43mScalability Intro – Shyam Jeedigunta, Google & Bob Wise, AWS (Any Skill Level)
28mPractical Approaches To Scaling Distributed Transactions