Default Gateways

[Ed Molinari]

To all,
I have a TCP/IP routing question that was spurred from something I saw on a client site and can't quite make sense of and am hoping some kind 'guru' can explain it to me.
Exhibit A: node with IP 135.46.152.10 subnetted 255.255.0.0 gateway 135.170.10.1
How does this work( and by the way it does ! ) ? I have been working for 2 years under the impression that the gateway must reside on the same subnet as the host and this is clearly not the case in the above example. The only thing that I can conceive of to make this work is that the router's table for that port is subnetted as 255.0.0.0, and that there are multiple subnets on that port, but that would defeat the purpose of subnetting because the whole 135 network would be a single collision domain.
Can someone please explain how this works and why a network might be configured as such?
If more info is needed I would be happy to discuss this more so I can figure out what is going on here.

Thanks in Advance,
Ed Molinari

[Luis Martinez]

Ed,

Please see comments below.


------------
Ed Molinari at 3/30/99 10:16:47 PM

To all,
I have a TCP/IP routing question that was spurred from something I saw on a client site and can't quite make sense of and am hoping some kind 'guru' can explain it to me.
Exhibit A: node with IP 135.46.152.10 subnetted 255.255.0.0 gateway 135.170.10.1
How does this work( and by the way it does ! ) ? I have been working for 2 years under the impression that the gateway must reside on the same subnet as the host and this is clearly not the case in the above example. The only thing that I can conceive of to make this work is that the router's table for that port is subnetted as 255.0.0.0, and that there are multiple subnets on that port, but that would defeat the purpose of subnetting because the whole 135 network would be a single collision domain.

Comment - We currently have a scenario that closely resembles this. You are correct in your assumption that the route table(s) on the router(s) is subnetted differently. With most routers you can "super-net" and/or create a custom routing table that will allow traffic from hosts on different subnets and route traffic to those specific subnets through filtering and minimize the traffic on each subnet in your organization. Although this may not be the case here with your clients network. Routers that are very good at doing this are Cisco routers. This type of setup can get very involved unless you know exactly what you are doing. Most organizations do go this route unless they have a routing expert available to them and they have the bandwidth and router to handle the traffic generated.

Comment - Something to try: Traceroute the paths the packets take to and from the GW. This will start to give you an idea as to which routers are doing the filtering/custom routing.

Can someone please explain how this works and why a network might be configured as such?
If more info is needed I would be happy to discuss this more so I can figure out what is going on here.

Thanks in Advance,
Ed Molinari

[Ed Molinari]

Thanks Luis for your response,
I have never before worked in large enterprise newtworks like the one here which is a major NJ telecommunications company. My experience with LAN/WAN was on 500 user networks and working with Proxy NATs, XYLAN OmniSwitch VLANs, RRAS, and 2524 Ciscos. So it comes of no surprise to me that there is a reason for not only making this work but why it is desirable in large networks.
My somewhat textbook view that I have implemented in the past is obviously not the only way the world works. I just have to find some good material that I can study more about this.
One more question, is this super-netting anything like what the ISPs use to implement CIDR in order to keep their routing tables at a minimum?


Thanks again for your response,
Ed Molinari


------------
Luis Martinez at 4/1/99 10:18:52 AM

Ed,

Please see comments below.


------------
Ed Molinari at 3/30/99 10:16:47 PM

To all,
I have a TCP/IP routing question that was spurred from something I saw on a client site and can't quite make sense of and am hoping some kind 'guru' can explain it to me.
Exhibit A: node with IP 135.46.152.10 subnetted 255.255.0.0 gateway 135.170.10.1
How does this work( and by the way it does ! ) ? I have been working for 2 years under the impression that the gateway must reside on the same subnet as the host and this is clearly not the case in the above example. The only thing that I can conceive of to make this work is that the router's table for that port is subnetted as 255.0.0.0, and that there are multiple subnets on that port, but that would defeat the purpose of subnetting because the whole 135 network would be a single collision domain.

Comment - We currently have a scenario that closely resembles this. You are correct in your assumption that the route table(s) on the router(s) is subnetted differently. With most routers you can "super-net" and/or create a custom routing table that will allow traffic from hosts on different subnets and route traffic to those specific subnets through filtering and minimize the traffic on each subnet in your organization. Although this may not be the case here with your clients network. Routers that are very good at doing this are Cisco routers. This type of setup can get very involved unless you know exactly what you are doing. Most organizations do go this route unless they have a routing expert available to them and they have the bandwidth and router to handle the traffic generated.

Comment - Something to try: Traceroute the paths the packets take to and from the GW. This will start to give you an idea as to which routers are doing the filtering/custom routing.

Can someone please explain how this works and why a network might be configured as such?
If more info is needed I would be happy to discuss this more so I can figure out what is going on here.

Thanks in Advance,
Ed Molinari

[Luis Martinez]

Ed,

It is the same as CIDR that ISP's use. If you want to know more on this subject and advanced routing, you can check out "Advanced IP Routing With Cisco Networks". The book is from Cisco Press (surprise). Have fun.

-Luis


------------
Ed Molinari at 4/1/99 12:12:04 PM

Thanks Luis for your response,
I have never before worked in large enterprise newtworks like the one here which is a major NJ telecommunications company. My experience with LAN/WAN was on 500 user networks and working with Proxy NATs, XYLAN OmniSwitch VLANs, RRAS, and 2524 Ciscos. So it comes of no surprise to me that there is a reason for not only making this work but why it is desirable in large networks.
My somewhat textbook view that I have implemented in the past is obviously not the only way the world works. I just have to find some good material that I can study more about this.
One more question, is this super-netting anything like what the ISPs use to implement CIDR in order to keep their routing tables at a minimum?


Thanks again for your response,
Ed Molinari


------------
Luis Martinez at 4/1/99 10:18:52 AM

Ed,

Please see comments below.


------------
Ed Molinari at 3/30/99 10:16:47 PM

To all,
I have a TCP/IP routing question that was spurred from something I saw on a client site and can't quite make sense of and am hoping some kind 'guru' can explain it to me.
Exhibit A: node with IP 135.46.152.10 subnetted 255.255.0.0 gateway 135.170.10.1
How does this work( and by the way it does ! ) ? I have been working for 2 years under the impression that the gateway must reside on the same subnet as the host and this is clearly not the case in the above example. The only thing that I can conceive of to make this work is that the router's table for that port is subnetted as 255.0.0.0, and that there are multiple subnets on that port, but that would defeat the purpose of subnetting because the whole 135 network would be a single collision domain.

Comment - We currently have a scenario that closely resembles this. You are correct in your assumption that the route table(s) on the router(s) is subnetted differently. With most routers you can "super-net" and/or create a custom routing table that will allow traffic from hosts on different subnets and route traffic to those specific subnets through filtering and minimize the traffic on each subnet in your organization. Although this may not be the case here with your clients network. Routers that are very good at doing this are Cisco routers. This type of setup can get very involved unless you know exactly what you are doing. Most organizations do go this route unless they have a routing expert available to them and they have the bandwidth and router to handle the traffic generated.

Comment - Something to try: Traceroute the paths the packets take to and from the GW. This will start to give you an idea as to which routers are doing the filtering/custom routing.

Can someone please explain how this works and why a network might be configured as such?
If more info is needed I would be happy to discuss this more so I can figure out what is going on here.

Thanks in Advance,
Ed Molinari

[Brent Jones]

I'm having problems with a PC that will not connect to certain sites on the internet. I suspect router table corruption. What can I do?


------------
Luis Martinez at 4/1/99 10:18:52 AM

Ed,

Please see comments below.


------------
Ed Molinari at 3/30/99 10:16:47 PM

To all,
I have a TCP/IP routing question that was spurred from something I saw on a client site and can't quite make sense of and am hoping some kind 'guru' can explain it to me.
Exhibit A: node with IP 135.46.152.10 subnetted 255.255.0.0 gateway 135.170.10.1
How does this work( and by the way it does ! ) ? I have been working for 2 years under the impression that the gateway must reside on the same subnet as the host and this is clearly not the case in the above example. The only thing that I can conceive of to make this work is that the router's table for that port is subnetted as 255.0.0.0, and that there are multiple subnets on that port, but that would defeat the purpose of subnetting because the whole 135 network would be a single collision domain.

Comment - We currently have a scenario that closely resembles this. You are correct in your assumption that the route table(s) on the router(s) is subnetted differently. With most routers you can "super-net" and/or create a custom routing table that will allow traffic from hosts on different subnets and route traffic to those specific subnets through filtering and minimize the traffic on each subnet in your organization. Although this may not be the case here with your clients network. Routers that are very good at doing this are Cisco routers. This type of setup can get very involved unless you know exactly what you are doing. Most organizations do go this route unless they have a routing expert available to them and they have the bandwidth and router to handle the traffic generated.

Comment - Something to try: Traceroute the paths the packets take to and from the GW. This will start to give you an idea as to which routers are doing the filtering/custom routing.

Can someone please explain how this works and why a network might be configured as such?
If more info is needed I would be happy to discuss this more so I can figure out what is going on here.

Thanks in Advance,
Ed Molinari