Query related to HSRP and RSTP

[Subin Kallidumpil]

Hi Team,

I need help on the above attached scenario.

Rapid Spanning tree protocol is running between Distribution layer and Access Layer.

Distribution Switch -1 is root switch for Vlan 2,4,6,8 and the hsrp interfaces for the same vlan is also active on the same distribution switch-1. Distribution Switch 2 is acting in standby mode for these vlans

Similarly, Distribution Switch -2 is root switch for Vlan 3,5,7,9 and the hsrp  interfaces for the same vlan is also active on the same distribution  switch-2. Distribution Switch 1 is acting in standby mode for these vlans

At the core, only core switch-1 is actively forwarding the data and core switch 2 is in standby mode.

My task is to replace the Distribution switch-2 from the network and replace it with the another switch which is same in terms of hardware.

Would like to know whether this activity can done without involving any downtime.

I was thinking of making some configuration changes to spanning tree priority and hsrp priority for vlans 3,5,7,9 on distribution switch-2  in order to reduce their priority  on distribution switch-2 and which will make them active on distribtuion switch-1.

After this i can remove the Distribution switch-2 from network and replace it with a switch of same configuration (Software and hardware)

However i am not sure if my methodology will cause any other downtime or complication and have more adverse effect.

Regards

K Subin

[Abzal]

Hi,

You can simply make distribution 1 STP primary for 2,4,6,8 VLANs and secondary for 3,5,7,9 VLANs. The opposite thing for distribution 2 STP primary for 3,5,7,9 VLANs and secondary for 2,4,6,8 VLANs. Then once you remove distribution 2 from the network distribution will be root bridge for all those VLANs.

Distribution 1:

spanning-tree vlan 2,4,6,8 root primary      // priority will be 8192.

spanning-tree vlan 3,5,7,9 root secondary  // priority 16384

Distribution 2:

spanning-tree vlan 2,4,6,8 root secondary

spanning-tree vlan 3,5,7,9 root primary

Make sure that access switches have lower priority than both dsitribution switches(8192,16384).

Hope it will help.

Best regards,
Abzal

[Subin Kallidumpil]

Azbal,

Currently Distribution1 is already root for vlans 2,4,6,8 and secondary for 3,5,7,9 VLANs. Similarly distribution 2 STP primary for 3,5,7,9 VLANs and secondary for 2,4,6,8 VLANs.

However the moment i remove the distribution 2 there will a downtime of 30 seconds (Time taken for RSTP convergence and time to detect that hsrp memeber has failed) after which distribution 1 will become active for all vlans.

I want to get over with this small down time. That's what i am trying to do.

Regards

K Subin

[Abzal]

Hi,

If your access switches also in RSTP mode and they connected to both distributions as per diagram. Then I believe RSTP converge very fast less than second. But with HSRP timers default is 10 sec. (holdtime).

standby timers

in vlan 10

...

standby 10 timers 1 4

In this way Hellos will be send every 1 second and Holdtime is 4 sec. If after 4 second Hello is not received then HSRP peer is down. So second peer will take over.

Hope it will help.

Best regards,
Abzal

[Subin Kallidumpil]

Hi Abzal,

Thanks for the information. This will help a lot.

I have some more query realted to same topic.

Can you please provide some information/steps which i should take into consideration when i would be placing the temporary distribution switch into the network.

Since i would not want that when i place the switch into the networkit causes any further downtime. However i would surely like to have this switch become active for the vlans it is supposed to be.

Regards

K Subin

[Abzal]

Hi,

I think this design guide can help you

http://www.cisco.com/en/US/docs/solutions/Enterprise/Campus/HA_campus_DG/hacampusdg.html

Hope it will help.

Best regards,
Abzal

[Subin Kallidumpil]

Abzal,

This doc provides a design methodology.

However i was looking out for steps from hands on configuration perspective.

I would be able to configure the temporary switch with out any issues. However when i will connect this temporary switch to the network, at that point what all steps are required to have a smooth migration.

I am actually looking out for that information.

Subin

[jeevak mukadam]

Hello Subin,

Abzal is right,Rstp provide faster convergence compair to taditional 802.1d STP process.Rstp ALT which is backup to root port that keep track of all alternative path to the root bridge but hsrp take time to converge.

Make sure that all the switches in the network operate in the transparent mode so any configration errors will not cause changes in the other switches in the network.

[Abzal]

Hi,

I summarized best practices and recommendations:

1. Make sure that all access and distr switches are running in Rapid-PVST+(RSTP) mode.

2. Distr1 need to be STP primary root for 2,4,6,8 VLANs and secondary root for 3,5,7,9 VLANs. The same for multiple HSRP groups.

3. Check and make sure that there is no access switch with lower bridge priority than primary

and secondary roots among access switches.

4. L2 links between access switches and Distr1 switch must be in a trunk mode and allowing proper VLANs range(2-9).

5. If you have configured Native VLAN it must be the same on both ends.

6. Configure access ports where hosts/servers/printers etc. are connected with Spanning-Tree PortFast feature.

7. Configure trunk ports with "switchport nonegotiate" to avoid DTP negotiation this can save seconds.

8. I would recommend to use VTP mode transparent. It's not easy to do it if you have lots of devices. But

   it will help you in situations by mistaken deleted or resetted VLAN database on VTP server.

9. If you have configured EtherChannel link aggregation configure in static mode turn off dynamic negotiation.

10. Tune HSRP preempt delay to avoid black holes(if you have it).

11. Ensure that the distribution node has connectivity to the core before it preempts its HSRP standby

    peer so that traffic is not dropped while connectivity to the core is established.

12. You can configure aggressive HSRP with millisecond timers, just before replacement and put it back for faster convergence.

    Example: standby 1 timers msec 200 msec 750

13. Depending on your network traffic rate you can configure L3 EtherChannel link between Core and Distributions.

14. Make sure Distribution using both links going to Core1/Core2. CEF will take care of load-balancing. CEF must be turned on(by default it is actually).

15. Check HSRP preempt if it is turned on.

16. You optionally tune STP by turining on Root Guard on designated port on root bridge.

17. The recommended best practice is to measure the system boot time, and set the HSRP preempt delay

    statement to 50 percent greater than this value. 

18. Provide an L2/L3 link between the two distribution switches to avoid unexpected traffic paths and multiple convergence events.

    To be independent from access switches. This can introduce unexpected behavior in the event of a failure.

Hope it will help.

Best regards,
Abzal

[Subin Kallidumpil]

Hi Azbal,

Thanks a lot for providindg all this information.

This will surely help me conducting the activity.

Regards

K Subin

[Abzal]

Hi Subin,

Glad to help. Mark as asnwered if you satisfied with the answers. It'd help for somebody else who will have the same questions.

Hope it will help.

Best regards,
Abzal

[jeevak mukadam]

ITs very helpful,

Thanks ABZAL

[paul driver]

jeevak,

Please don't forget to rate Abzal  posts.

res

Paul

[jeevak mukadam]

Noted with thanks.