subnetting question

[garyrivers]

so, i have 4 switches.  one switch feeds the other 3 switches that are in different locations.

i have config-ed a vlan 500 with an IP of 10.30.13.0/28. enable EIGRP, etc. and attached a port to that vlan. 

On the remote switches, i'd like to have

the remote switches use the same IP range, only with a smaller subnet mask.

thoughts?

Thanks!!

[Joseph W. Doherty]

Your other "switches" are L3 switches?

Generally you cannot route with overlapping networks, which is what you want to do?

[garyrivers]

yeah, they're 3850's.

well crap! i was hoping to be able to do that.

this is how i have it now:

I only have one source pipe and i have to feed the 3 remote sites.

currently, i have the 3 sites being fed via 3 interfaces on the top switch. and each port has its own IP range...i have attached the ranges and ports. what i'd like to have one port feed the remotes.

so, how would you config this scenario?

[Joseph W. Doherty]

I don't fully understand what your limitations are. For example, unclear on your limitations of ports and/or IP usage.

If your remote sites just have one link, and very small IP usage, would it be possible to make them strictly L2 and host the IPs on your "core" L3 switch? That would expand the IPs you would have for hosts using the /28.

[Martin L]

To have all devices on the same IP range you will need full mesh topology;  everyone would share 10.30.13.0/28 subnet.  Otherwise, Hub-spoke topology uses /30 links but each hub-spoke pair uses individual link and subnet.

Maybe there is a way to "cheat" one into another.

by "have one port feed the remotes" you mean like a router os a stick? router with sub-interfaces?

Regards, ML
**Please Rate All Helpful Responses **

[garyrivers]

actually, i misspoke! there are 3 interfaces, 1 to each remote switch.  BUT the 3 ports are in vlan 500.

[Martin L]

this may work with L3 switches;  add interface vlan 500, assign IP on the same subnet,  check if interface vlan 500 is UP UP, make sure all switches are doing trunking on hub-spoke link and vlan 500 is allowed on trunk.

enable EIGRP for network that belongs to vlan 500.  Since switches are doing trunking, spoke-spoke and hub-spoke should be established (spoke eigrp hellos will go via hub).  all spoke to spoke traffic will go via hub as well. you may need to disable split horizon on hub.

Regards, ML
**Please Rate All Helpful Responses **

[Richard Burts]

The original post describes a main (top) switch and 3 subsidiary switches with a single vlan 500. But in a follow up post there is a png attachment that shows vlans 200, 201, 202, and 203. Where do these vlans come from and what do they do?

The png attachment also shows a layout of IP addresses starting at 10.30.13.0 and incrementing through 10.30.13.15, suggesting 4 IP addresses per site. That is a bit problematic, especially if they intend to treat the subsidiary switches as layer 3 switches.

There are probably several ways that this network could be implemented:

- one option would be to treat the subsidiary switches as layer 2 only switches connecting to the top switch. The top switch would operate as layer 3 and would do all routing for the network. The subsidiary switches might have just a single vlan connecting to the main switch (with all the switches sharing the single vlan) which sounds like one possibility suggested in the original post having just vlan 500. Or perhaps each subsidiary switch might have its own vlan (suggesting a network with multiple subnets).

- another option would be to treat each switch as a layer 3 switch. The network might use vlan 500 as the management vlan where each switch had an interface in vlan 500. And each switch could have its own data vlan and could run a routing protocol to advertise its data vlan over vlan 500 so that it was reachable from other switches.

The layout of IP addresses from 10.30.13.0 to 10.30.13.15 is probably problematic. If the network is to be implemented with all switches sharing a single vlan (and all routing done by the top switch) then this address plan might work. But if each switch is to operate independently then each block of addresses needs subnetting logic where each subnet has a network address, a broadcast address, and user addresses and the given addressing plan needs to be changed. 

Perhaps the original poster can provide some clarification about how the network is intended to work?