March NSFNET T1 Usage by Service
Olivier, The use of NNStat for statistics collection on the T1 backbone has not been disabled at any point of entry. Ultimately, drastically differing in and out counts are not unusual. On average, 27% of the packets and 50% of the bytes are from ftp: Jan 1992 28% packets 50% bytes Feb 1992 27% packets 50% bytes Mar 1992 29% packets 52% bytes Using the net primarily for ftp will appear as a small amount of traffic into the backbone and a significantly larger traffic volume out. Washington University at St. Louis (MO) exhibits a similar statistical measurement: Pkts (millions) Bytes (billions) IN OUT IN OUT Dec 1991 294 157 126 12 Jan 1992 332 212 139 17 Feb 1992 338 214 142 17 Perennially, they are the 1st or 2nd highest volume net on the NSFNET, apparently due to a very large PC archive accessed primarily via ftp. Similar numbers are noted with other high volume nets. It's not surprising that other nets will show the reverse of this IN/OUT volume relationship. We hope this information helps explain your concerns. Please let us know if we might provide further clarification. With sincere apologies for the delay in sending this information on to you, and thanks to Susan Horvath for her analysis and review, Jo Ann Ward Merit/NSFNET Information Services ********************************************************************************** Message: 12625724, 33 lines Imported: 12:01pm EST, Wed Apr 1/92 Subject: Re: March NSFNET T1 Usage by Service To: Merit/NSFNET Information Services-Interested Parties, Susan M. Horvath From: MARTIN%CEARN.BITNET@pucc.Princeton.EDU Merit/NSFNET.Information.Services-Interested.Parties@um.cc.umich.edu, Douglas.E.Van.Houweling@um.cc.umich.edu, nsfnet-reports@merit.edu, nsfnet-info@merit.edu cc: Elise Gerich <epg@merit.edu>, Jessica Yu <JYY@MERIT.EDU>, "Stefan Fassbender (GMD/EASInet)" <stf@easi.net> In-Reply-To: Message of Wed, 1 Apr 92 10:49:34 EST from <Susan.M.Horvath@um.cc.umich.edu> Thank you for your very interesting tarffic reports, there is a question which I have been trying to get answered without much success for some time, which is the following: Why is the traffic from ASN 590 (EASINet) into the T1 backbone so small whereas the traffic to ASN 590 is so high (ratio 1/1000 in february). (To be more precise: 78.736 Megabytes In, 82.14 Gigabytes Out)! My personal explanation is that you only take into account the incoming traffic and that for some reason, accounting has been disabled at the EASINET entry into the T1 backbone, therefore only the traffic coming via alternate access points (i.e. secondary or ternary) is really accounted for, however, I could never get any confirmation of this "theory". If this theory happened to be right, I would very much appreciate that adequate mention to it be made inside the files containing the traffic statistics. Olivier
Which just goes to illustrate that traffic-based cost-sharing models for [intercontinental] links are complicated by things like FTP where the origin address sends huge amounts of data for which it did not initiate the session. Some countries/entities turn out to be (grossly) net exporters of data. --SG
-------- ] From: Steve Goldstein--Ph 202-357-9717 <sgoldste@cise.cise.nsf.gov> ] Subject: Re: March NSFNET T1 Usage by Service ] Date: Wed, 01 Apr 92 14:41:12 EST ] ] Which just goes to illustrate that traffic-based cost-sharing models for ] [intercontinental] links are complicated by things like FTP where the origin ] address sends huge amounts of data for which it did not initiate the ] session. Some countries/entities turn out to be (grossly) net exporters ] of data. --SG As transport costs decrease, it is likely that there will be a shift towards information based billing (i.e. billing for access to archives, information servers, etc.) It's interesting to consider the impact that such a shift would have on today's network service providers. Networks that exist primarily to provide access to other organizations' data may have a difficult transition ahead. /John
participants (3)
-
jaw
-
John Curran
-
Steve Goldstein--Ph 202-357-9717