Question

Extreme X460-G2-48x-10GE4 Ring Topology with CISCO Switch 9600 (Virtual Stackwise)

  • 12 May 2020
  • 5 replies
  • 350 views

Hi Tech People, 

 

In our LAN Topology , we have 5 Exteme Switches in ring toplogy. There is single  2x10G link from Extreme switches  to two Core Switch paird in Virtual chassis e.g. Extreme Switch # 1 has single link 2x10G to the CORE and Extreme switch # 5 to the Core. 

I want to confirm that Can we make a loop free design in the case?  Suggest me the best design  as i m new to extreme network. Extreme switches are not in stack. 

thanks in advance.  


5 replies

Userlevel 6
Badge

Hi Rehman,

 

I don’t have much experience in this kind of scenario myself, but in general EAPS ring can have 3rd party devices between two Extreme switches, as long as these devices pass through EAPS control VLAN and data vlan to the ring ports of EAPS-enabled switches. Please have in mind that with third party switches, if something between these breaks, Extreme switches may not spot this failure, so second type of failure detection comes into play - failtimer expiry. It might make the failover time a bit longer compared to Extreme-only ring when link failure information is immediately sent to the Master node.

You may want to refer to the EXOS User Guide to better understand EAPS ‘under the hood’, or please let us know what else would you like to know. :)

https://www.extremenetworks.com/support/documentation/extremexos-30-6/

 

Hope that helps,

Tomasz

Thanks Tomaz,

 

 I would like to ask how the Cisco device understand the EAPS packet ? Spanning-tree and EPRS has different algoritham. EAPS will  running inside Extreme Switches and for the interfacing with Cisco switches, Spanning-tree is the only option. you can see the toplogy for better undestadning and can guide better. 

 

 

Userlevel 6
Badge

Hi Rehman,

 

Cisco does not support EAPS I believe. So it cannot participate in failover like Extreme switches (by generating Links Down packet if they detect their ring-participating port went down). But it can pass the control VLAN traffic for EAPS through. And actually, only EAPS Master node (in a single ring scenario) is responsible for any action (that is, unblock secondary port for protected data VLANs if the ring is broken; otherwise it is blocked to prevent broadcast storms).

I think it’s okay to try out. At least on some dummy control/protected VLANs besides current STP config (if I understood correctly that’s we way it runs right now?).

Good to mention, specify Master node as one somewhere in the middle of your access ring. As Master’s secondary port is to be blocked, it would be good to have rather similar distances, so that far-bottom or far-top switch’s traffic doesn’t have to go all way through other EXOS switches to reach the core.

If possible, I’d go with 2x10G in all ring links if the amount of traffic may require this (check current utilization).

 

Hope that helps,

Tomasz

Thanks for your suggestion, 

i want one clarification more that if we can run EAPS and Spanning-tree on  far-top and far-botton extreme switches switch simuntanously. e.g

 

Cisco Core = MSTP

Extreme switches= EAPS and MSTP on the top and bottom Switches (change Spanning Tree port priority) 

 

thanks for your support 

 

Userlevel 3
Badge

don’t do this. It’s a bad idea to combine EAPS and STP on one switch.

Cisco switches will normaly forward eaps control pakets. You even can configure EXOS switches that EAPS is working fine with 3rd party switches in ring.

The big challange is to trigger a arp/fdb flush in cisco switches.

 

But in my eyes using EAPS or STP with a blocked link is not solution, it a help and not state of the art anymore.

You should think about MLAG or Stack-Designs and on cisco side about stacking, bonding or vpc

Reply