Table of Contents
Lessons from Internet Services: ACID vs. BASE Dr. Eric A. Brewer UC Berkeley Inktomi Corporation
Our Perspective
Global Architecture
Search Engine as a Database
Other Problems
No DBMS...
ACID vs. BASE
ACID vs. BASE
The CAP Theorem (new?)
Forfeit Partitions
Forfeit Availability
Forfeit Consistency
These Tradeoffs are Real
PPT Slide
Traditional Architecture: Big SMP
Inktomi Architecture
Incremental & Absolute Scalability
Principles
Mission-Critical Attitude
Expect and Tolerate Failures
Fault Tolerance Options
Use Smart Clients
Server Pollution
Delivering High Availability
Conclusions
HotBot Awards
History
Cost/Performance
Architecture Comparison
Search Engine Goals
Relevancy of Results
The Web Will Reach a Billion Pages
New Data Coming Online is Relevant
Inktomi Database (simplified)
Basic Query:
Basic Query:
Parallel Query
|