25: Internet Measurement Issues and Tools
"Measure what is measurable, and make measurable what is not so." - Galileo Galilei"Far better an approximate answer to the right question, which is often vague, than an exact answer to the wrong question, which can always be made precise." - John Tukey"Statistics is prediction." - Edward Deming. Lecture outline: how to make sound measurements of the Internet?
1. Internet measurement basics
Continuous evolution of the Internet (from ARPANET to the modern era)
Why perform measurement?
What should be measured? Infrastructure, traffic and applications.
Performance metrics: Availability, loss, delay, and utilization.
Challenges: a) volume of data; b) need for time synchronization; c) privacy/ trust issues; d) sticky statistical properties (wild randomness)
2. Internet measurement techniques
Passive measurements:
Logs from server/ equipment; Counter based measurement (SNMP); Syslog
Packet capture (BPF, libpcap, tcpdump, winpcap, wireshark)
Flow capture (Netflow)
Active measurements:
Ping, Traceroute; Bandwidth measurement.
3. Some measurement projects
Active measurement infrastructures: RIPE-TTM, SURVEYOR
IEPM: Pinger project
CAIDA: Skitter and Scamper projects
IP network monitoring frameworks: AT&T and SPRINT
Primary reference for this lecture:
“Strategies for Sound Internet Measurement” by Vern Paxson [link]
Secondary references for this lecture:
“Internet Measurement: Infrastructure, Traffic and Applications” by Crovella and Krishnamurthy (1st chapter).