Gigabit Network

[Garion]

That subnet mask allows for 65,000+ addresses on a single network. That probably also means your whole dorm is probably on a single network and any broadcasts that is sent out by ANY box on your dorm is sent everywhere. Windows boxes are very, very chatty. Baaaaad news. Most of us try to cap the number of users per subnet at about 100, to minimize broadcasts and ease troubleshooting. Sheesh. Sounds like your school needs a decent network engineer!

Edit: Assuming that they have a single DHCP scope that you're part of, you are machine #1348. (5*255+73). Law of average says you're halfway through the scope. I can't imagine that they have that many PC's on a single subnet - Your dorm isn't that big! Either they started their DHCP scope somewhere other than the .0.x address or there's a bunch of dorms on one monster flat subnet. It's amazing it even works. Wonder how big the forwarding table is in the switches and if it's run out yet?

- G

 

[spidey07]

Just imagine the broadcast rate the router has to keep up with.

Hey Garion! Up for a little uni work?

 

[MikeDub83]

I can tell you that the DHCP server does not start at .0.x. I forgot what exactly the range is, but I don't think thats the problem.

There was a definite broadcast problem. That was one of the main reasons for purchasing the new switches that allow them to block broadcasts. I'm not sure if the broadcast problem still exists.

 

[spidey07]

fyi - switches INCREASE broadcasts. Routers block broadcasts. If there was a broadcast problem before, then adding the switches just made it worse. But from the IP addressing info you've posted I can imagine they had serious broadcast and arp issues.

Sorry to be so blunt, but from the sound of it your campus network is fvcked. In professional terms we call it a "melt down".

-edit- just for fun download ethereal from the internet and startup a trace. See what your broadcast frames and broadcast frames per second are. Then look at the trace, my money is on a ton of ARP and NETBIOS/NBT traffic.

 

[MikeDub83]

Yeah I would definately agree the network is definately screwed.

I'm trying to find out what type of switches were installed for you. These new switches that have the filtering capability replaced our old plain jane managed switches.

They explained that the purpose for their install was to block certain protocols. On of the reasons was to block someone from running a DHCP server. The other reason was to stop Windows SMB broadcasts.

 

[MikeDub83]

Using Ethereal... I ran the test for two minutes and received 567 packets, 77Kbytes.

27.5% Netbios Datagram Services - Microsoft Windows Broswer Protocol
11% Spanning-Tree Protocol
10% Internetwork Packet eXchange
16% Address Resolution Protocol
2.5% NetBIOS
9% Internet Group Management Protocol
10% Service Location Protocol


I have no idea what most of these are. How does it look to you?

 

[Garion]

There must be something else going on, since that's really not many broadcasts. The Admins might be smarter than we thought!

- G

 

[MikeDub83]

The packet loss is not happening randomly. It occurs in patches every 5-15 minutes. Any ideas?

 

[alrox]

Best way to get to the bottom of this is mail your network dept. with logs about the packet loss. They should get a competent person on the job and see if anything can be done.

 

[MikeDub83]

Thats what I'm trying to do. There is actually only one network administrator. He is not completely convinced that it's his end but is starting to realise that maybe he is at fault.

Thanks to eveyone who helped out.
- Mike Dubreuil
Sophomore of Computer Engineering
WIT, Boston, MA

 

[MikeDub83]

I was able to find out what type of switches they are. They are made by Enterasys and feature the Netsight Policy Manager. Any idea what could be up?

BTW: My net admin is finally looking into the problem.