Frame Relay Cloud Setup

[r-remien]

I have a Frame relay network with 14 remote sites (PVCs). The frame is a full T1 link. 12 sites have a 16 Kb CIR, 1 with 64 Kb and 1 with a 128 Kb CIR. The 12 sites are all burstable anywhere from 64Kb to 256 Kb. The hub router for the frame network is a 3640 with no subinterfaces. The cloud is all on one subnet. All sites come through the 3640 for LAN access between the sites and Internet access through the 3640 to the gateway router. There is a lot of slowdown and congestion currently. If I change all the pvcs to use subinterfaces and have all differenet subnets using VLSM, will I see any improvement in speed? Or are subinterfaces just an administrative advantage? Also, when I change all the sites to use different subnets, will I need to notify the frame relay provider about any changes I make?

Thanks,

RJ

[MickPhelps]

1) You won't have to notify your frame provider of anything that doesn't affect provisioning. Changing an interface on your router from point-to-multipoint to point-to-point is not a privisioning change.

2) You won't realize gains over your frame, as a rule of thumb. Your routing protocol may add overhead though.

3) It is an easier way to administer frame (p2p vs p2m), and more stable. No problems with inverse-arp, frame-relay maps, etc...

You need to determine what traffic slows down and why. Is it frame congestion (BECNs, FECNs & DEs) or e2e congestion (port speed not high enough)? Are you getting any errors on your interfaces or queue drops? Have you trended the usage to determine if your pipes are sufficient to support the number of users at each site?

Mick.

[r-remien]

I am seeing a lot of congestion.

1)What is e2e congestion(port speed not high enough)? How do you configure port speed? Also, I will not be running a routing protocol because I have several old routers in my frame that could not handle anything but RIPv1.

2) What about the response below: Are broadcasts sent over a frame relay network?

3) As far as bandwidth usage, I am using mrtg to monitor bandwidth. Although, I do not know how accurate the results are. I had one day where there were 5 spikes at the max 128 Kb. Although, all my users said they had problems all day long. As a rule, what percentage of the pipe needs to be consumed where there is a noticeable slow down?

Thanks,

RJ

[gary]

You would definately benefit from using subinterfaces, and also subnetting the WAN links, due to having the cloud as all one subnet you have one large broadcast domain, this in istself will cause the network to slowdown as packets will be broadcast to all sites.

I am running a frame relay network with 22 sites, all my remote sites are on separate subnets, and then I have the WAN links separately subnetted, using EIGRP I can advertise the networks across the the WAN.

So for example. (Remote Site)

E0 is 192.168.48.1 255.255.255.0

SO is 192.168.1.6 255.255.255.252

no ip directed-broadcast

encapsulation frame-relay

no fair-queue

frame-relay interface-dlci 101

frame-relay lmi-type ansi

eigrp 1

network 192.168.48.0

network 192.168.1.0

Central Site

interface Hssi3/0.1 point-to-point

description Frame Link to Anywhere

bandwidth 128000

ip address 192.168.1.5 255.255.255.252

ip directed-broadcast

no ip mroute-cache

frame-relay interface-dlci 106

eigrp 1

network 192.168.1.0

+ network address for ethernet.

You do not need to notify your frame relay provider, how you run your circuits is up to you.

Regards,

Gary Jennings

Network Mgr

Abacus Polar Group