Multiple subnets on same segment with Cisco 2600 - performance problem

[Aslan67]

Good afternoon!

I'm new here so be gentle.

I have a nice, new, certified CAT6 LAN with a problem.

Here is some history on our network. We have a class B WAN which has been further subdivided by our issuing authority into smaller masks due to the fact that most locations don't have 256 hosts and we have too many locations for whole subnets.

Our location has 4 masked subnets (4 different sets of addresses XXX.XXX.YYY.XXX, where YYY is different on each subnet). None of these is a whole 256 host mask. All of these subnets ride the same physical network where a Cisco 2600 (single ethernet port) connects us upstream. We have appropriate subnet masks to define them, i.e. 255.255.255.192 or 224, etc.

Our issuing authority saw fit to provide secondary gateway addresses within the Cisco so that each subnet had it's own virtual gateway.

Now, one particular subnet, let's call it 026 for sake of example, is where a server resides, providing resources for whatever host is part of the NT/2000 domain. Any host configured for the 026 subnet has no major network performance issues (outside of intermittent stuff probably created by hosts on the other subnets) when accessing resources on that server. None of the subnets have any performance issues when accessing internet or upstream resources. Any host placed on a subnet other than 026 experiences some of the most unGodly network performance problems when accessing server shares (drives, printers).

Now, I had given the server an auxilliary IP address on one of the other subnets and tried again, but still there was a performance issue.

I recently found an article on WildPackets that explains how the RFC has changed from an old doctrine to a new doctrine that defines the way a router can redirect ARP traffic. RFC 1812 says that redirects should be discarded if a new gateway is more than one hop (which it shouldn't be) but that a Redirect Message must NOT be generated in a router unless all the following conditions are met:
The packet is being forwarded out the same physical interface from which is was received.
The IP source address in the packet is on the same Logical IP (sub)network as the next-hop IP address, and
The packet does not contain an IP source route option.

It seems to me that we are failing on the last 2 points. There is a reference that either the IP stack has to support Redirects or (as a stop-gap) Proxy ARP must be enabled in the router so as to prevent unnecesary traffic loading, performance degradation and propagation delays.

I have no access to the router to check its config and our authority is less than helpful on this situation. I'm trying to gain ammunition to set a course of action on this problem.

What I want to know is this: Is a Cisco 2600 too old to support Redirects, and if so, could this problem be attributed to Proxy ARP being disabled or am I barking up the wrong tree entirely?

Remember, I've pretty much ruled out the possibility of cabling being the issue as when a host is experiencing this problem on a different subnet as 026, I can put it back on the 026 subnet and eliminate the problem.

Bottom line is, I have to use the other subnets as our authority can't give me a larger, complete subnet.

Any ideas from anyone?

Thanks,
Brett

 

[baddos]

2600 supports redirects, but it can be turned off by the no ip redirects command.

A 2600 is way to wimpy a router for you to vlan your network with though. You should get a 3550 switch to handle most of your traffic.

 

[Aslan67]

Thanks baddos. I actually have a Cisco 4000 configured to break up my subnets into VLAN's via Enterasys switches, but I have been reluctant to do so since I have not exactly found the root cause of the problem. I'm sure implementing this router, as well as making my server multi-homed for the subnet(s) that require its resources, will fix the problem but I'm still not sure *what* the problem is.

I'd like to know if anyone out there has seen this scenario and whether the RFC information I have found is relevant or not.

Thanks!
Aslan67

 

[sharkyj72]

I dont see how ip redirects are relevent here.

Also a 2600 is fine for routing between VLAN's.

Never use proxy arp unless you've got a really old device that doesnt have a d/f gateway configured in its routing table.

1. Make sure all devices on each subnet have their gateway configured correctly.

2. Check to see if your using a tagging protocol between vlans (802.1Q, ISL etc) if so then you should be able to switch between the vlans using tagging (configure sub-interface on the 2600).

3. Try using the ip route-cache same-interface command.


See the following,

http://www.cisco.com/en/US/partner/...ation_example09186a00800949fd.shtml#cisco2600

for an example.

 

[Degg]

Here is another idea. Is your router a 2611? If it is, then its only going to be 10Mb. If you are doing inter-Vlan routing between 4 Vlans, and you have a lot of traffic on each Vlan, your router may be over worked. Check the CPU usage on the router, and also check the usage of the interface. That would explain why hosts on the same subnet of the server have no issues, because they dont go to the router. Other subnets must be routed, and the router cant handle the amount of traffic. If you get a router with a FE interface, that should solve the problem (your 4000). Degg
Network Administrator

 

[vammu]

You can try

ip route-cache same-interface


which will maintain a route cache between the subnets.

Make sure when you do
show ip int e0
...
IP autonomous switching is enabled
IP autonomous switching on the same interface is enabled

 

[Aslan67]

Thanks to everyone for their suggestions on my issue. I won't be able to change the router settings myself as our authority is the only one with access to it. I believe, after hearing all your suggestions and the research I've done, the best route for us to take will be to actually segregate our subnets into VLAN's and use the Cisco 4000 so everyone gets their own router port.

You've all been a great help!

Aslan.