Cornell Implementing Bandwidth Charges

Sabalon writes "Cornell University is planning on implementing a plan where if faculty, staff or students use more than 2GB of bandwidth a month, they will be charged for the additional bandwidth usage. The article mentions that last year over 100,000GB worth of files were sent from Cornell's network. I'm sure this is not the only school doing this or moving to this. I'm sure the conspiracy theory people will see this as a suggestion by Microsoft to stop students from getting those pesky Linux iso images. At least, according to the RIAA, CD sales around Cornell should now skyrocket :)" It'll be interesting to see how this plays out. Since students often have accounts on several different university machines, I suspect the more rebellious ones will be running an assortment of proxies and redirections to get around the restrictions.

Comments

Necessary, but stifling

[Templar]

I agree that this is absolutely necessary, as I pay the bandwidth bills at my company and know what it's like, but they have to be careful not to stifle innovation, as the security features they will now need become more and more complicated.

What will this do to the thousands of students that use 802.11b at the library and other campus buildings? Will the charges be based on MAC address? Since MAC addresses are so easy to spoof, authentication will become necessary. How can that be done easily across multiple platforms?

The new measures might wind up costing them more than they expected. How about limiting speed by user? That would not get in the way of most legitimate research, but it would render P2P movie sharing useless.

Re:Necessary, but stifling

[CerebusUS]

Will the charges be based on MAC address? Since MAC addresses are so easy to spoof, authentication will become necessary. How can that be done easily across multiple platforms?

I get really really tired of people who don't read the article before posting. I had mod points and decided to write this instead.

From the article:
"The last -- and most debated -- charge is a new Internet-use fee, which some officials refer to as the "pay by the drink" plan. The fee will be based on the bandwidth consumption associated with a specific network address, known as an IP number. Every computer on a network has a unique IP number."

Points off for michael as well. billing by the ip address means that in order to proxy a connection without the router seeing it, you'll have to locate the proxy on the same network and then THAT IP address will see its usage shoot up.

Re:Necessary, but stifling

[mkldev]

they have to be careful not to stifle innovation

Good point. When I was in grad school, I ran ftp.mklinux.org out of my dorm room. The IT dept. grumbled a lot about it, but since it was academic research, they let it go. We were noted as serving 20 gigs in a single day on occasion, with typical being between 1 and 3.

If they had set up a similar policy, even if they only charged 1/1000th of a cent per megabyte, it would have cost me on the order of $150/month. At that rate, it's roughly half the cost of leasing your own T1, complete with paying an ISP to service it.

On the other end, if they charged a half cent per meg., my continuous 384 kbps would cost about as much as a full OC3.

Put another way, this is how -not- to keep serious CS students at your university, guys....

Not that new..

[Gortbusters.org]

I have a friend at Vanderbilt, he has a 200 meg per day quota. If he exceeds that quota he'll get a warning the first time, and the second time he will loose his LAN connection.

I have heard other stories as well where they have monthly quotas and then get charged - or more often - service revoked.

They're running an ACADEMIC network

[b.foster]

Contrary to popular belief (and, yes, contrary to my own usage patterns in college), Universities provide network resources for academic uses. This usually means that they intend for those networks to be used for research (this is often the main reason the institution exists), completing assignments, and communicating with one's professors and peers. These networks are not and never have been intended to be used for entertainment purposes.

Cornell's change is a Good Thing(tm) in that they will encourage private entities to provide metered, regulated internet service to the members of the campus community. In this way, the individual members, not the aggregate, will be responsible for paying for the proportion of resources they use. Because, after all, when everybody agrees to divide the check, most of the people at the table order lobster. It's time for the liberals at universities to drop their Ivory Tower facade and face the fact that human nature is a greedy algorithm.

Good to see Cornell implement a sensible policy

[shalunov]

I've long advocated usage-based billing as the way to manage campus bandwidth (see slide 6 of `QoS Appliances Considered Harmful' presentation at the spring 2002 Internet2 member meeting).

If you think you're entitled to use as much network capacity for as long as you want because you already pay tuition, compare network use to printer use. No-one expects to be able to print 10000 pages a day, day after day, on the department printer for free. This is because it is understood that each page costs something. The marginal cost of transit of each packet on Internet1 is non-zero: universities are billed for traffic.

Internet2 traffic is a different matter: the marginal cost of transit of a packet is zero, and there's plenty of capacity to play with.

Re:Ugh.

[ayf6]

Probrably because the school doesnt WANT to reach their peak bandwith. They dont just get bandwith for free... They have to pay an upstream provider just like the majority of the rest of us. There are very few of use that are fortunate to be a Tier-1 provider. You school probrably - i have no idea of what their actually agreement w/ their telcom is - is that they pay for X bandwith but a rise to X' will cost them money. They have X' prime bandwith to use however they pay for X and have to pay an additional fee when they rise to X'. For example you may have a OC3/T3 line put into your company but have it capped at 25Mb/s but if you have need to rise to 45Mb/s you can call your telcom and ask them to do this. This is perhaps the reason your friend in the NOC thought you had more bandwith then you really did. "Sure joe we have and OC3 here..." but he neglects/doesnt know that its only a partial OC3.

Re:2GB??!

[MrLint]

Remeber that a 56k modem has a theoretical maximum of 17G/30 days,

Idiotic

[riptalon]

University officials sent out letters to researchers -- including those who, for example, move around large amounts of sky-telescope data -- to warn them of the billing changes. The university offered "to round off the sharp edges" for researchers who will be adversely affected.

They better had! The assumption that high bandwidth use is all down to music filesharing and other "non-work related" activities is not necessarily well founded. I work for a different large US university and regularly need to transfer data from the other US universities or europe to analyse. I can get transfer rate of 250-750k per second depending on the time of day. This translates to very roughly 1-2 Gb per hour and I might spend all data selecting datasets and leave the transfers going all night and maybe the next day too, to get what I need. A transfer of upto 100 Gb over a couple of days followed by a month or more to analyse the data (before I need more) is not unheard of. A 2 Gb per month limit would stop my research in its tracks and there must be people at Cornell that need similar bandwidth to me, for their work.

This sounds more like a money making scheme than a real problem. Universities usually get charged a fixed amount for their external connections, whether they use them or not. If they have maxed out their connection and everyones transfer rates are sufferring then slapping quotas on the undergrads, who don't do any work and so shouldn't need large amounts of bandwidth, is the answer. Charging users is just money grabbing since the money isn't going to go to add more bandwidth, since the demand for bandwidth will have fallen when the charges are intoduced.