lecture21: Input data
The lecture concerns the problem of getting the input data
for network optimization: primarily the traffic matrices used in
many problems.
Topics covered
-
- Planning Tasks
- Planning Cycle
- Traffic Matrices
- Packet traces
- Packet traces issues
- Packet trace info
- {\tt tcpdump output}
- Packet trace example
- Packet traces pros
- Packet traces cons
- Reducing the data size
- Netflow
- Netflow records
- Timeouts
- Netflow records
- Netflow example
- Netflow example application
- Example traffic matrix computation
- Netflow pros
- Netflow cons
%
- Sampled Netflow
%
- Netflow records
%
- Impact of sampling
%
- Example distortion
%
- Netflow samples
%
- Netflow records
%
- Impact of sampling (flows)
- SNMP
- SNMP data collection
- Irregular sampling
- SNMP pros
- SNMP cons
- Example SNMP data
%
- Other sources
- The Problem
- Who put the CAT in CATscan?
- Performance Tomography
- History
%
- Using a Traffic Matrix
%
- Traffic matrix estimation
- Problem formulation
- Solution
- The catch
- Inverse problems
- Tomogravity~
- Discussion