Stefano, The only concern I have is that some of the information requested will have privacy impact. Is it possible to draft some sort of confidentiality agreement between the sharing parties, this project, and Apache to make sure this information is kept private and only used for the project?
Also, would it make sense to remove the last 8 or 16 bits from the IP address? I think it would also be a good idea to round the timestamp to the nearest day. Thanks, Reza -----Original Message----- From: Stefano Andreani [mailto:[email protected]] Sent: Tuesday, October 09, 2012 3:34 PM To: [email protected] Subject: DDR update procedure Hi all, In order to setup a process to build and maintain the DDR, the first requirement is to identify a way to allow the uploading of http logs from contributors, in order to analyze the user agents. We should not receive just user agent lists, but full http logs, for these reasons: - we need the timestamp for each user agent, in order to identify the frequency of each user agent (we will not be able to process each user agent, but hopefully enough to cover 99% of the http requests; - we need the source IP address, in order to have a geographical MAP of distribution of the Devices, for two reasons: (1) analyzing IP addresses (using a geographical DB, like GeoIP) we would have a map of uncovered regions, so we would be able to improve the global coverage chasing contributions from specific regions. (2) the same device can have different user agents depending on the region where is has been commercialized and using IPs we can improve analysis. At the same time, uploading such information without a clear policy about how that data is handled could imply privacy issues, so we must keep the upload area private and guarantee that the information is used consistently with the objectives of DeviceMAP project, and not for other purposes. Do you agree on this? Anyone from the Apache infrastructure team can help to identify what is the technical solution to satisfy these requirements? Cheers, Stefano.
