myk2mic:
Beta technician
Firms.use.external.servers to.handle.dhcp.that I mentioned Windows.is.not.strange .the.idea is.to.limit.the services that router performs.to .allow.it.perform.its.core functions.which is routing.I .have encountered such setup so don't.limit urself.to.only.Cisco.device.

Yes bt.like.I stated it is used to shelve load on.the.router so as to allow.router.perform.it's.core.function which is.routing.only.small.Firms.with.small users can.afford.to.load dhcp services on.the.router. U.won't try it in a.large firm .talking.from.experience

True. However processing dhcp requests barely spike up router cpu processor these days with the heavy artillery of ram, cpu and asics pumped into routers.
Reason is more admin related than performance.

Franklyn4:
I built an ether channel and enabled IP-ROUTE , where my core and distribution layers are L3 switches. pls how do i make the ports of my core interfacing my distribution switches to be in passive mode . the idea is that i dont want my distribution layer to participate in my routing.

your question doesn't quite string together as i am not sure its a question on design and/or etherchannel. however the following might help:

1) if by making ports on core to be in passive mode you are referring to etherchannel then you should use the lacp protocol to negotiate bonding as it is the one of the 2 methods (other being pagp) with the passive mode

2) note that if the other end of the bundled ports are not put in the lacp mode of active then the etherchannel would not come up.

3) if by you do not want your distribution layer to participate in routing you mean it should operate at layer 2 and not have any form of routing process enabled then you need to configure the ports facing the access switch southbound and core switch northbound to be trunk ports which goes for the etherchannel configured. However this is generally deemed bad design because you are expanding the scope of your broadcast domain and a single host in an access switch with vlans forwarding on all distribution switch when it sends out broadcasts would be received on all layer 2 switches forwarding same vlan and hosts assigned to same vlan, this would ultimately eat your bandwidth and bump up your processing overhead which is not good.

4) if by you do not want your distribution layer to participate in routing you mean the distribution terminates all the vlans from the access layer southbound and then routes northbound to the core then this is good design and is okay. if you are using more than one distribution layer device for redundancy with a normal switch you would then need to run an fhrp like hsrp, vrrp or glbp to allow for redundancy so if one switch goes off the the hosts have another to use as gateway with the virtual ip of the fhrp = the ip default gateway gotten from the dhcp pool sent to the hosts, if you have smarter boxes like the cat4500 and cat6500 then you do not need fhrp as the 2 cat4500/6500 would operate as one single box with their svi being the default gateway for corresponding vlans.

Hope this all helps and isnt too fuzzy