Frame DLCI help

[djjazzyjeff]

I have a basic question for you all that I can't seem to get answered to my satisfaction. My company has a frame pvc, CIR of 1024kb between their two sites on the east and west coasts. I need to understand the logic that dictates the choice of DLCIs at each end. Currently I have defined DLCI 105 at one end and 100 at the other. Are the DLCI choices mine to make or the provider's (AT$T)? Do they have to be different or can/should they be the same? Would mismatched DLCIs have an effect on performance? Thx in advance......Jeff

 

[Ntr0P]

DLCI's are assigned by the provider and should be different on each end of the PVC. Be sure to use the DLCI they assign if you are hard coding them into your router. If you use the wrong DLCI's performance will be impacted in that there will be no communication.

 

[djjazzyjeff]

I have hard coded them into my routers, is an alternative to not do this and what implications/benefits are there?

 

[Ntr0P]

Provided Inverse-ARP is possible (and I believe it is on AT&T), you do not need to hard code them, they will be learned. The dynamic solution makes life a little easier, but if you like the control and don't want to rely Inverse-ARP, you can hard code them. I do not have mine hard coded and have not had any issues.

I am assuming you are using Cisco routers as well.

 

[djjazzyjeff]

Yeah, 2620 on my end and 2611 on the other. The reason I'm looking into this is end-user complaints about performance across the circuit. It is lengthy, running from the Boston area to Vancouver, BC. I have MRTG running and have used ttcp for measurement purposes, at midday ttcp shows 112kbs between 2 test hosts with normal traffic and domain replication running. The complaints seem to be all related to network browsing using 2k machines. Users in Windows Explorer drill down through shares on windows boxes and each drill down drags. I have many Solaris 2.8 hosts but no complaints with these. There are a number of Sol machines in Vanc running Samba, I haven't used that myself but wonder about excessive traffic generated by these. What do you think, I'm getting hammered harder each day and it's starting to hurt.....Jeff

 

[Ntr0P]

Doesn't sound like it has anything to do with your FR. If it is a network browsing issue, I would break out a sniffer and try to figure out where the packets are being held up. Another thought, are you running any kind of QoS that might be causing an issue?

 

[djjazzyjeff]

No QoS on this end to blame this on. I guess I'll have to sniff to get at the source of this. Can you recommend a sniffer that I can use across the WAN?

 

[Ntr0P]

If you want something that will truly work across your WAN, VisiNet has a nice product. A bit pricey though. You may want to talk to AT&T and see what they can do for you. I know they have a product as well. It is a fee based service, but maybe they can help you.

As for a plain ol' vanilla packet sniffer - I'm partial to Observer. You could always throw the packet sniffer on workstations on each end of the WAN and determine where the issue lies. If it isn't at either LAN, then start looking at the WAN. I imagine you will find the issue on the LAN somewhere though.