Excellent public studies of Internet traffic by protocol exist from NSFNet times up to 2002 or 2003; we thank Odlyzko, Claffy et al for their work.
However, has anyone else noticed a serious lack of data after the end of the studies summarised in “Longitudinal study of Internet traffic 1998-2003”, Fomenkov, Keys, Moore & Claffy, 2003? (This has been discussed in the past, I think.) All that seems to be available is vendor handwaving of the sort Andy Odlyzko takes so much trouble to debunk; and that without any source data.
Curiously, no-one seems to be funding this stuff in Europe, either.
While KC is obviously best person to answer, my understanding is that with the disapearence of ATT inside of SBC and the disappearance of MCIUUNET inside of Verizon all their traffic that CAIDA used to get was pulled. CAIDA gets it no longer because now it is proprietary. IMO a bad situation
The pendelum swings back and forth. At the moment, companies have been getting bashed when they make an effort to make data available and then researchers spend their efforts to de-anonymize the data.
Companies still spend a lot of money and hire research companies to analyze the data privately; its just not worth the hassle of releasing
it publically or giving it to academics for free.
You might find some additional data of interest here:
Granted, not quite full of data the repository aims to have at this
point, but perhaps what is there you didn't know about may be of
interest and that perhaps my pointing to it encourages people to
contribute to the project (note, not just anyone can gain access to
all the data).