This is a multi-part message in MIME format. --------------CBC6F8CA1772C24E49EB2652 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit --------------CBC6F8CA1772C24E49EB2652 Content-Type: message/rfc822 Content-Transfer-Encoding: 7bit Content-Disposition: inline X-Mozilla-Status2: 00000000 Message-ID: <38BDAA87.2A2CC282@pilot.msu.edu> Date: Wed, 01 Mar 2000 18:40:55 -0500 From: Jason Justman <justmanj@pilot.msu.edu> X-Mailer: Mozilla 4.61 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 To: Jeremy Bowers <bowersj2@msu.edu> Subject: Re: Napster at MSU? References: <87zosiflt7.fsf@pfaffben.user.msu.edu> <38BD879A.315969FF@crushedice.com> <006b01bf83c6$07452b00$46ab0a23@user.msu.edu> Content-Type: multipart/alternative; boundary="------------1A0B56A9C7CF4349C7E382FB" --------------1A0B56A9C7CF4349C7E382FB Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Jeremy Bowers wrote: > I would like you to look at this page: > http://mrtg.cl.msu.edu/mrtg/cc2-gw.2.html > It's a traffic analysis for a router. This router happens to be a special > router on campus, as I understand it, this is our Merit link. In other > words, all outgoing and incoming traffic goes through this router. > yes, you are connect.. all outbound world traffic goes over this primary link to merit networks. > I want you to read the page _backwards_. Look at the year graph first. The > graph is almost flat... an odd spike in June I can't explain, but overall, > the router averages ~1/4 maximum capacity with a roughly equal amount of > data incoming and outgoing. A small spike on the incoming data in Aug/Sept > when the students return to campus... then all hell breaks loose in the third > week of October. I cannot confirm that Napster was released around that time. all hell breaking lose is not an additional 10Mbits/s.. hell, that could be all the kiddies x-fering rh6.1 iso.. who cares what its from? not me.. it should be msu's responsability to have the infrastructre that is capable of handing high bandwidth.. [Image] first, the odd spike in june was from a large DOS attack (well, medium sized) originating from MSU from a few unsecured sun boxes.. just enough to flood a t1 or so.. > I _can_ confirm that the most recent Napster beta was released on January > 20th... give it a bit of time for word to travel on campus, and we're just > n time for the Febuary spike that has not ended since. from doug -- We have two OC-3 links (155 Mbps full-duplex each) from MSU to the Internet, but only a single FDDI (100 Mbps, half-duplex) from campus to the Merit routers that attach to these OC-3's. This will be upgraded by the end of the week. -- feb 7th im not specifically sure if the upgrade has been in place, but i think that merit just sucks for external conectivitiy anywasy.. > Since that spike > has occurred, network connectivity has been shot to _hell_ around here. The > MSU web page doesn't even _load_ sometimes, images can be dropped, and other > such wonderful things, and thats _internal_. External connectivity is even > worse; my personal non-MSU sites will sometimes just cease to exist for a > period of a minute or two, and that's happenning multiple times per hour. I > had a telnet session to pilot _die_ about 3 hours ago; not just hang forever > waiting for input, which sometimes happens when pilot goes on the fritz, the > connection _died_. 2 minutes later, it was fine. I couldn't connect before > then, though. > What you are forgetting is this: --doug he biggest recent problem is that the Computer Center gigabit router has been running out of CPU. This shows up as packet loss almost everywhere. Foundry's tech support came up with a memory optimization change that improved this. With luck, that will resolve the packet loss difficulties here. --feb7th This is noticable in the http://mrtg.cl.msu.edu/netmon/msu/alldown.html page.. I suggest you take a look at it before blaming internal MSU connectivity on napster, rather than some major issues in the campus rings.. as of right now, there are 359 alerts on routers dropping packets > Now, some points (again, correct me where I go wrong): > > - If the recent network problems were solely due to routing issues with > Merit, then we should expect to see network graphs looking like last > Saturdays... low, flat, non-activity as nobody can get in or out. Instead, > we see traffic skyrocketting nearly all of last month. I am concluding that > we have a traffic issue. > > - I can't _prove_ it is Napster, all I'm saying is that based on the > experiences of other Universities, it is a likely explanation. If it ISN'T > Napster, I _still_ want to know what it is. > its because the GigE implementation wasn't tested on a production bandwith level.. > - If... if if if it is Napster, then I want it banned or throttled > immediately. Let the students whine about censorship; let them first look up > the definition of censorship before they do so, because banning Napster > ain't censorship by any stretch of the imagination. I've got projects that > are suffering because the network is being toasted. stop using napster as a blame. Its not saturating off campus links, MSU is experiecnign some upgrading issues.. let doug and pals figure it out.. or better yet, call CIC and ask why so many hosts are losing packets.. > Now, I'd like to ask Doug about these things directly, and ask if he's maybe > afraid of student backlash or something. Me, I just want the damned network > to function like it's supposed to. read msu misc.. he's not.. --from doug Sure. In general, you can whatever you want with your connection, as long as it's legal, you follow the AUP, and any other relevant university rules and regulations. There are no plans that I'm aware of to either limit bandwidth or restrict access to any sites. We always reserve the right to temporarily cut off external or internal sites if the traffic interferes with the network as a whole, but that generally is applied to hacking or denial-of-service incidents, not general bandwidth hogs. So.... Make sure your MP3's are legal, and have at it. --feb29th > Are there any obvious flaws, other then the fact that I don't have the tools > to prove or disprove that it is Napster's fault? obvious flaws - local campus network sucks - not because its oversaturated (unless you see a link from the fddi ring thats aggregate bandwidht (inbound+outbound)=100 (cuz its half duplex)), but because something (probably still gige), isnt behaving nicely with the rest of the fddi rings.. j --------------1A0B56A9C7CF4349C7E382FB Content-Type: multipart/related; boundary="------------722D293DA84B8D59E9E13532" --------------722D293DA84B8D59E9E13532 Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit <!doctype html public "-//w3c//dtd html 4.0 transitional//en"> Jeremy Bowers wrote:I would like you to look at this page:yes, you are connect.. all outbound world traffic goes over this primary link to merit networks.
http://mrtg.cl.msu.edu/mrtg/cc2-gw.2.html
It's a traffic analysis for a router. This router happens to be a special
router on campus, as I understand it, this is our Merit link. In other
words, all outgoing and incoming traffic goes through this router.
I want you to read the page _backwards_. Look at the year graph first. Theall hell breaking lose is not an additional 10Mbits/s.. hell, that could be all the kiddies x-fering rh6.1 iso.. who cares what its from? not me.. it should be msu's responsability to have the infrastructre that is capable of handing high bandwidth..
graph is almost flat... an odd spike in June I can't explain, but overall,
the router averages ~1/4 maximum capacity with a roughly equal amount of
data incoming and outgoing. A small spike on the incoming data in Aug/Sept when the students return to campus... then all hell breaks loose in the third week of October. I cannot confirm that Napster was released around that time.
first, the odd spike in june was from a large DOS attack (well, medium sized) originating from MSU from a few unsecured sun boxes.. just enough to flood a t1 or so..
I _can_ confirm that the most recent Napster beta was released on Januaryfrom doug --
20th... give it a bit of time for word to travel on campus, and we're just
n time for the Febuary spike that has not ended since.
We have two OC-3 links (155 Mbps full-duplex each)
from MSU to the Internet, but only a single FDDI (100 Mbps, half-duplex)
from campus to the Merit routers that attach to these OC-3's. This
will be upgraded by the end of the week.
-- feb 7thim not specifically sure if the upgrade has been in place, but i think that merit just sucks for external conectivitiy anywasy..
Since that spikeWhat you are forgetting is this:
has occurred, network connectivity has been shot to _hell_ around here. The
MSU web page doesn't even _load_ sometimes, images can be dropped, and other
such wonderful things, and thats _internal_. External connectivity is even
worse; my personal non-MSU sites will sometimes just cease to exist for a
period of a minute or two, and that's happenning multiple times per hour. I
had a telnet session to pilot _die_ about 3 hours ago; not just hang forever
waiting for input, which sometimes happens when pilot goes on the fritz, the
connection _died_. 2 minutes later, it was fine. I couldn't connect before
then, though.
--doug
he biggest recent problem is that the Computer Center gigabit
router has been running out of CPU. This shows up as packet loss
almost everywhere. Foundry's tech support came up with a memory
optimization change that improved this. With luck, that will
resolve the packet loss difficulties here.
--feb7thThis is noticable in the http://mrtg.cl.msu.edu/netmon/msu/alldown.html page.. I suggest you take a look at it before blaming internal MSU connectivity on napster, rather than some major issues in the campus rings.. as of right now, there are 359 alerts on routers dropping packets
Now, some points (again, correct me where I go wrong):its because the GigE implementation wasn't tested on a production bandwith level..- If the recent network problems were solely due to routing issues with
Merit, then we should expect to see network graphs looking like last
Saturdays... low, flat, non-activity as nobody can get in or out. Instead,
we see traffic skyrocketting nearly all of last month. I am concluding that
we have a traffic issue.- I can't _prove_ it is Napster, all I'm saying is that based on the
experiences of other Universities, it is a likely explanation. If it ISN'T
Napster, I _still_ want to know what it is.
- If... if if if it is Napster, then I want it banned or throttledstop using napster as a blame. Its not saturating off campus links, MSU is experiecnign some upgrading issues.. let doug and pals figure it out.. or better yet, call CIC and ask why so many hosts are losing packets..
immediately. Let the students whine about censorship; let them first look up
the definition of censorship before they do so, because banning Napster
ain't censorship by any stretch of the imagination. I've got projects that
are suffering because the network is being toasted.Now, I'd like to ask Doug about these things directly, and ask if he's mayberead msu misc.. he's not..
afraid of student backlash or something. Me, I just want the damned network
to function like it's supposed to.
--from dougSure. In general, you can whatever you want with your connection, as long
as it's legal, you follow the AUP, and any other relevant university rules
and regulations. There are no plans that I'm aware of to either limit
bandwidth or restrict access to any sites. We always reserve the right to
temporarily cut off external or internal sites if the traffic interferes
with the network as a whole, but that generally is applied to hacking or
denial-of-service incidents, not general bandwidth hogs.So.... Make sure your MP3's are legal, and have at it.
--feb29thAre there any obvious flaws, other then the fact that I don't have the toolsobvious flaws - local campus network sucks - not because its oversaturated (unless you see a link from the fddi ring thats aggregate bandwidht (inbound+outbound)=100 (cuz its half duplex)), but because something (probably still gige), isnt behaving nicely with the rest of the fddi rings..
to prove or disprove that it is Napster's fault?j