Transcript Overall Architecture

Overall Architecture
Participants
George Brett
Dan Eklund
Russ Hobby
Bruce Lowekamp
Kevin Walsh
Important Points
• No central coordination
– but may be hierarchy
• Both current and post-mortem analysis
• Need privilege/access levels
Overall Architecture
Data Collection
Trouble Reporting
Data Repository
Analysis
Data Repository
•
Distributed database
–
•
•
We are assuming ability to receive and
transmit between other areas
Archiving
Forwarding queries?
1. Does not forward
2. Forwards
–
needs delegation of authority
Analysis
• Threshold detection
• Diagnosis
• Display
– Might be a separate area of Display & UI
Threshold Detection
•
•
•
generates events - producer
generates trouble tickets
generated by
1. probe/host-based agent
2. DB analysis
3. application
Diagnosis
•
•
•
•
•
Meaning mining
Handles trouble tickets
Requests data from repository
Triggers new tests
Results go to trouble reporting
Trouble Reporting
• Creates trouble tickets
• Not all trouble is equal
– agent created
– human created
– privilege important here
• Creates solved case and records fix
– “knowledge history”
Overall Architecture
Data Collection
Data Repository
Threshold detection
Diagnosis
Application
Trouble Reporting
Display/UI
Overall Architecture
Need to agree on
–
–
–
–
Data representation
Producer interface
Consumer interface
Location service