Anon101 wrote: yes. unfortunately, at the moment that is a satisfactory speed.
It actually got so slow today that it signed me out of MSN, like dialup did 4 years ago...
yea...its doing that for me too atm :/
I dont know if anyone else has noticed, but they are having major networking problems in the learning grid at the moment, I can't help but think all the problems are linked.
Its completely different, and on a different network. The pipe from midman to warwick uni uses different bandwidth than the pipe to warwick resnet.
Why, they have to lie about it is still beyond me, I'm definately heading to this focus group thing on wednesday, does anyone else go to that?
Please do. I plan to publish something against accommodation early next week.
My main concerns at the moment are:
How are they detecting these so called P2P Protocols, "protocal" is a very general term, are they monitoring use of known P2P ports? because if they are thats a very poor method of detection. Or can they detect that a certain user is flitting between different servers in a P2P like fashion? Or are they just assuming anyone using a SOCKS5 is using P2P (which again would be very poor).
The traffic shaper does this job. Initial package inspection is an intelligent combination or traffic signature and port. If the packet is undefined then a deep package inspection is performed to find out the useage. Its a cat and mouse game with BT and major networking companies... each time a new signature is made the companies need to update shaper software. This is why *some* p2p packets may get through, as they havnt been updated on the shaper software as yet.
Since theyve banned P2P protocals, (including IRC etc even though they deny it) I dont see how people can use P2P in active mode anyway.
Read above for how p2p could work. As for IRC the block will be lifted imminently. [when the network has settled a little] If you would like to use it before then use a bouncer. If you pm me I'll set one up for you. Or if you connect via vpn on the uni network it will work.
Whatever they are using to measure, it's a very unreliable method, because Rootes, and Cryfields are not even mentioned. erm, Rootes, has gotta have the largest filesharing community.
The only basic detection method is through deeppackets on the shaper. It will only flag up the worst offenders.
How often are they updating these lists that they are restricting bandwith to? or is it realtime, does it restrict and unrestrict depending on if any "protocol" is in use in a realtime fashion.
The resnet is now in a set of subnets. Each subnet has certain bandwidth. A new subnet is currently being setup for offenders. Once it is done offenders will be moved there and severely throttled and the subnets restricted will be unrestricted. As of now, no subnets are restricted.
Why can't we access an up to date list, because surely it is our right to know if our bandwith that we are paying foris being restricted.
Email ITS and ask them to notify you if your subnet is restricted?
Why can't they restrict induvidual users bandwith? It seems there restricting the bandwith at the network switches in induvidual halls.
Thats what they are working on now as explained before
Why do they need to "investigate" which users are using these "p2p protocols" surely that have a logging system, in which case they either know, or don't know. It's black and white.
Investigation consists of getting that information from the shaper.
Basically, there are so many holes in the email, and it's so incredibly vauge. Maybe I'm speculating but this is what I THINK:
My best guess is they are either assuming people are using P2P if they are connecting to a number of servers and flitting between them (i.e connecting to multiple sources for a file, Bittorrent esq.) AND/OR they are assuming that SOCKS5 is a "P2P Protocal".
I think they were trying to keep it simple so non technically minded people would understand
I'm also guessing, that they have not actually mapped the IP's to room numbers in halls, I reckon when they plugged in the switches they did it pretty much randomly, and therefore, have no idea what data is being sent from what room. Hence them having to restrict bandwith to particular halls. My circumstancial evidence for this is someone at the opposite side of the house has an IP which is only 1 number away from my address. If they had done it surely, as room numbers increase, the IP numbers would increase 1 by 1.
This is totally incorrect. You are not understanding the way DHCP works. Basically upon connection your port in your room will show as being connected to x IP. DHCP exists due to the number of IP's
. Your IP will change anyway when you get a new lease
Hence why they are restricting whole halls, and why they need to "investigate" induvidual cases.
They can only provide and restrict by subnet not by user. If ITS or any other ISP were to divide pipes like that between users it wouldnt be good (you'd have 1.67kb/s bandwidth)
The standard ISP model works on contention ratios and thats how resnet works too. P2P is completely fucking up a lot of networking standards.
Still have no real suggestion for their rather random list though!
they're probably trying to start somewhere?
Anyway, what do other people think?
I personally think we're going in the right direction. I think that more work on the shaper is still required and the best way to go forwards is to reserve 10MBS of the pipe for HTTP, HTTPS, POP, MSN, and SSH at the highest priority.
I also think subnet throttling should work in affected subnets so they only have 400k or so bandwidth, and the other bandwidth reserved for these "standard" protocols.
It does need to improve but remember its only been inhouse for a few weeks.
I'm currently writing a motion against accommodation which includes a refund demand on this issue along with several other things. I'll take it to the union and see if i can whip up some support
And if ITS are reading this, an update would be greatly appreciated.
Well actually I prefer to be seen as "that undergrad hammered off his head from too much vodka" but its all good
Apologies if I'm not up to speed technically. I'm not a CS student.