There are distinguishable roles in establishing, running and extracting results from any set of data about networks. In some cases, one entity will handle more than one role. In all cases, the access and expertise of the entity impacts how they carry out the role. That is, the operator of a network is in a better position to understand the network being measured (access), although they may not have as much experiment and analysis expertise as a non-affiliated researcher. The 3 key roles are:
- Definer of the measurements/data collection
- Access: operator has better understanding of how the network fits together and potentially any “aberrations”
- Expertise: network expertise necessary for both measurement and collection
- Capturer of the data
- Access: operator can see more detail; might collect and summarize/sanitize before sharing; nonetheless, some data is visible outside of networks and can be captured by external parties
- Analyzer of the data
- Expertise: network expertise is essential for explaining the inevitable anomalies; data science expertise allows further inferences to be drawn for trending, expectations
- May be done by more than one entity at any given time, and they may have no relationship to the original data capturers
- May be done at the time of data capture, or some time later (days, months, or even years)
Since the analysis of data may occur at some considerable time later than its capture, there is always an additional element of variability in that the networks, operating practices, tools, and even the things that are being measured may change over time.
Finally, apart from the entities that cause the measurements/collection, there are users and uses of the measurements. Network operators may use measurements to identify issues in their network, researchers may use them to identify changes in the overall landscape of the Internet, and regulators may use them to create policies.
Think that sums it up? Way off the mark? E-mail comments to measuring<at>techark.org , please!