Question to the experts? Is this is a good practice or not?
Hi all ..... I have notice something in company's network:
we have applied cisco hierarchical Model.
All switches in access Layer are working as L2.
All switches in distribution layer are working as L3(doing routings), All of the are 6500 series.
2 switches in the core layer but here is the ISSUE ----> One switch is working AS L3 while other is working AS L2.
All distribution switches are connected to both core layer switches.
I personally don't know if this is a common practice or not. But i have been thinking what will happen if L3 core switch fails ?????? the other device is configured as L2. So it need some time to compute STP which means downtime.
Is this a common practice, if not !! what is the impact of such design?
I think when cisco implemented router operation in ASIC there is no need to let Core layer switches act as layer 2 devices.
PS: Please if you can point to some document for me to read regarding this issue I'll be great full.
Thanks in advance
we have applied cisco hierarchical Model.
All switches in access Layer are working as L2.
All switches in distribution layer are working as L3(doing routings), All of the are 6500 series.
2 switches in the core layer but here is the ISSUE ----> One switch is working AS L3 while other is working AS L2.
All distribution switches are connected to both core layer switches.
I personally don't know if this is a common practice or not. But i have been thinking what will happen if L3 core switch fails ?????? the other device is configured as L2. So it need some time to compute STP which means downtime.
Is this a common practice, if not !! what is the impact of such design?
I think when cisco implemented router operation in ASIC there is no need to let Core layer switches act as layer 2 devices.
PS: Please if you can point to some document for me to read regarding this issue I'll be great full.
Thanks in advance
. : | : . : | : .
Comments
-
dtlokee Member Posts: 2,378 ■■■■□□□□□□It's tough to make a full assessment of what you have and if it can be implemented in a better fashon. Having the access layer operate as layer 2 switches is not a problem. As for spanning-tree you can look at RSTP or MST as alternatives to PVST which can take 50 seconds to converge. If the distribution layer is doing all your routing (inter-vlan and what not) the core switches can be running as L2, and the one running as L3 might only ned this for connectivity to a larger network (other sites or the Internet for example). I would also look at some first hop redundancy protcols like HSRP or preferably GLBP to ensure your hosts on the access layer have some redundancy.
Just some random thoughts, more information would help more.The only easy day was yesterday! -
CCIE_2011 Member Posts: 134I may provide you with more information if you just mention it.
Here are some.
My companies network is a very big network.
we're using other switch blocks(i.e. server farm, enterprise farm, ... etc) for shared services
Q: Now if core is working at L2, and distribution working at L3. is the STP at core will include all devices in other layers ?. : | : . : | : . -
dtlokee Member Posts: 2,378 ■■■■□□□□□□If the interfaces from the distribution switches to the core are configured as routing interfaces then no, but if they are configured as trunk ports then yes. There are other ways to handle the core like QinQ protocol tunnels that would limit the impact of changes to the access layer across the core.The only easy day was yesterday!