How will new CIT pricing effect this site?

[DeltaOne81]

[Q]Precisely. 80% of the commodity bandwidth usage comes from ResNet, and the vast majority of that is peer-to-peer file sharing, and of that, about 85% is outgoing.[/Q]
So what was so wrong with the packet shaping they were doing? Giving low priority to the things like P2P and file sharing, so that legitimate traffic doesn't get slowed down, and these 'excess' uses only get bandwidth when there's bandwidth to spare.

Seems perfectly reasonable to me - no one at Cornell has a right to bitch about it; everything would still work, just slower; and since the excess services are what's using all the bandwidth, Cornell wouldn't feel obligated to pay for more . But no, that solution wouldn't require foistering more financial burden on families in poor economic times, so it can't be good.

[cmoberg]

As a former employee of CCS  (Cornell Computing Services, what CIT was known as when I worked there) I have a different bent on this situation.  Back then, the department existed as something of an GL hybrid, part overhead and part revenue.  One element of the latter, was leased line rental.  This consisted of 300 and 1200 baud (am I dating myself) direct connections to the mainframe at Langmuir Lab (what is it called now??)  Other Cornell departments rented their connections from and were billed by CCS (CCS held the contracts and paid the rental of the actual phone lines to NY TEL)

Todays communications landscape is very different but the operating model is likely the same.  My guess is that the overhead side of the house is getting squeezed. This leads to examination of fixed and variable costs.  While Cornell intranet traffic can be viewed to a certain extent as fixed cost,  the charges for traffic in and out of the campus net is based on utilization and is clearly a variable cost.

With the explosion of peer to peer file sharing, I am sure Cornell like other univerisities is grappling with issue of huge packet charges to outside service providers.  Many universities have taken a similar position to that proposed  by CIT.  Charging is an effective control on utilization (and one that has been used for years at Cornell)

Perhaps CIT will adopt the fixed price base, with ala carte overage common to the cell phone and commerical networking industry.    The fixed price base could of course be FREE.  

Chris

[CowbellGuy]

Chris Moberg wrote:

Many universities have taken a similar position to that proposed  by CIT.

Got examples?

[jeh25]

To be fair to CIT, this really is a case of a few abusers of the system making it bad for everyone else. To see what I'm talking about, checkout http://www.cit.cornell.edu/computer/students/bandwidth/charts.html

If the accounting costs don't kill you, it seems to me that charging per bit is the easiest way to prevent this tragedy of the commons from occuring.  From a personal freedom standpoint, I think I'd rather have Cornell charge per bit rather than filtering packets based on content.

[Will]

John, I looked at the link.  Interesting statistics.  Do you (or anyone else in the know) happen to know if CIT ever did anything about those top 20 users like, say, suspend their ResNet connections, or issue warnings or something?

Also, does anyone know what kind of hassle would be involved in just installing a firewall over ResNet preventing filesharing (at least outgoing filesharing) to begin with?

[DeltaOne81]

[Q]The fixed price base could of course be FREE.[/Q]
You do realize that the fixed price base service currently costs $45/month for ResNet, right?

[Q]To be fair to CIT, this really is a case of a few abusers of the system making it bad for everyone else.[/Q]
Absolutely. So why not just crack down on the abusers? I know some people who use their ResNet connection as servers. I've tried to talk them out of it. I agree it's ridiculous. The NOC/ResNet/CIT without a doubt have the ability to find out how much bandwidth everyone is using (how else would they charge based on bandwidth). Why can't they just add something to the Resnet clause for next year saying you're not allowed to use your connection as a server (including to server P2P) and then take a few hours and find the 30 people that use 1/3rd of Cornell's bandwidth and send them cease and desist notices.

You could argue that the new pricing scheme is essentially the same approach. And I'd agree it is IF and ONLY IF the threshold for extra charges is reasonably high - say, 15 or 30 gigs per month (this IS the same organization that set the limit at 27 gigs in 3 days just last year - 30 gigs in a month should be fine).

That would allow you to download plenty of movies/music/software update/etc and still be fine, yet at the same time, make outrageous users who serve multiple gigs per day pay dearly.

I get the bad feeling that the "threshold" will be like a gig or two, if there is one. I hope I'm wrong.

-Fred

P.S. I also think it should only go for ResNet, the idea of departments having to pay for their professors' and staff members' usage is downright stupid.

[cmoberg]

You do realize that the fixed price base service currently costs $45/month for ResNet, right?

No, I did not. Things have changed since my time in the dorms.  However, that charge is on par with a DSL or Cable connection in my local market. And like DSL/Cable there should be a level of performance comittment with the service.  

Chris

[jtwcornell91]

It seems like this sort of thing has happened in the past with disk usage, back when disk space was expensive.  Systems with a "customer" model have a fixed quota of storage space, while more "community" oriented systems send periodic emails to the biggest users to cut down when space began running out.  (slack.net used to have a file called /tmp/diskpigs with everyone's disk space usage.)

It seems like the high-volume tax on residental users is an appropriate disincentive (but with some warning sent when a user is getting close to having to pay extra) but departments should have the opportunity to lobby for an exemption if they really are doing something legitimate (like Grid Physics or Cornell-promoting webcams) rather than just swapping music at the office.

Incidentally, the same few people who are ruining this situation for everyone are also the people who helped bring on the current crisis in internet radio by getting the recording industry lathered up about people copying music over the internet.  So if you like we can blame them for CornellPass too... :-/

[DeltaOne81]

[Q]However, that charge is on par with a DSL or Cable connection in my local market. And like DSL/Cable there should be a level of performance comittment with the service.[/Q]
Well, yes and no. First, yes it's about the same price that people pay for cable or DSL connections. BUT, one household often splits the cost between people, making the price more like $10 to $20 per person per month. ResNet connections have no such option (well, technology-wise yes, but it's against policy and they *can* catch you).

Second, Cable/DSL has to set up people all around town, Cornell just has to run wires in a handful of closely located building. It *should* be cheaper for them to do so than it is for a broadband provider.

Third, Cable/DSL companes make a PROFIT at those prices, Cornell should be (and claims to be) charging us cost. If so, they have a poorly streamlined system (what is new?).

"One of the guys" - what you're suggesting is otherwise known as packet shaping. It can give lesser priority to some traffic over others. It can even entirely block certain kinds of traffic if desired. It's already in place. It's not ideal, they've had some problems, but nothing that can't be worked through.

JTW, I disagree. Actually, Napster ruined it. But, more appropriately, the record industry ruined it by attempting to attack their customers instead of finding a way that works best for everyone (kinda sounds like some athletics department we know, doesn't it? though even more extreme). That $1 billion Napster deal would have been the best thing for music since sliced bread (mmmm, mixed metaphor).

[cmoberg]

DeltaOne81, I agree that it should be cheaper for CIT to install the basic infrastructure.  However, bringing network infrastructure to old buildings is not without challenge.  As I recall most of west campus res life buildings lack raceways which suggest higher cost than a typical new office building.

Also, from what I remember of the various campus wiring projects during my tenure at CCS, the conduits around campus were not easily navigated (the original broadband plant was put in when I worked there) And Cornell does not allow over head wiring (AFAIK)

The split cost issue is not so cut and dry.  I have a friend who got nailed for hooking his intranet to the cable modem.  On the other hand, my DSL supplier endorsed my use of a DSL router and connection to my home/office lan.  So in my case, I do enjoy shared use of the DSL pipe.

What is the term of use that prevents you from piggy backing a lan (or proxy server) on the single connection.  Is it designed to keep room/suite mates from sharing a single port.

For your $45 what speed drop to you get in the room?  What sort of interconnect exists from the dorm to the campus backbone?  Is there any individual service comittment?

Chris

[CowbellGuy]

I don't see them charging more for electricity to old buildings. Oh wait, they don't charge for electricity at all...

[DeltaOne81]

[Q]However, bringing network infrastructure to old buildings is not without challenge.[/Q]
Given, but it's there now, so they shouldn't be impacting costs too much anymore.

[Q]The split cost issue is not so cut and dry.[/Q]
True, some companies allow it, some don't. How hard they enforce it is another story. In the Ithaca area though, they do allow it, so this is the alternative that Cornell is 'competing' with. Of course, ResNet is really a monopoly since DSL isn't possible in the dorms and Cable is expensive to have installed. If Cornell goes ahead with this plan with some large threshold, I wouldn't expect to find the number of people that stay in the dorms drop. Internet is virtually a utility by now (for college studnets at least). I woulda left had they done this earlier.

Btw, I found be surprised if CIT doesn't announce it until after the housing lottery, so people don't have a chance to consider it as a factor.

[Q]What is the term of use that prevents you from piggy backing a lan (or proxy server) on the single connection. Is it designed to keep room/suite mates from sharing a single port. [/Q]
* No Shared Connections: A network connection supplied by ResNet is solely for the use of the individual subscriber assigned to that connection. Connections may not be shared among multiple users. ResNet subscribers cannot use any mechanisms (either hardware or software) to provide network connectivity to non-subscribers. The use of hubs, hublets, wireless access points, and similar devices requires authorization by ResNet and is limited to personal use by the subscriber.

And yes it is. There are pretty much two reasons. One, it's not possible for Cornell to support every make of router out there, so if people expect help with trouble connections, they can't use it. Two, if a connection is registered to someone's netID (as each is), then they are responsible for anything that happens on it. If you let someone on your connection, and they, or anyone they let use their computer, does something against campus policy, the person who owns the subscription is liable. And, of course, there's money, which is reason #3 and probably the main one.

[Q]For your $45 what speed drop to you get in the room? What sort of interconnect exists from the dorm to the campus backbone? Is there any individual service comittment?[/Q]
a) I'll admit the connection is damn good. Quicker than cable or DSL. Everyone has a 10BaseT switch, though your outside connection won't even approach that, but a couple hundred K/sec is feasible if you're lucky and the stars align correctly.
b) I dunno.
c) What exactly do you mean by this? Bandwidth? tech support?

Okay, this is beginning to get long and detailed, which isnt good for a forum. But my original point remains. If the problem is a limited number of peole, and a limited number of uses, why not just take action against that... instead of against everyone?

Ah Cornell... sigh.

[Adam \'04]

One of my friends installed a wireless server in his dorm and supplied those in the know with free Internet. I don't think that person has been caught yet. ::laugh::

[jtwcornell91]

Adam '04 wrote:

One of my friends installed a wireless server in his dorm and supplied those in the know with free Internet. I don't think that person has been caught yet. ::laugh::

Has he been war-chalked? :-P