Switching

Can you explain RTG - redundant trunk group feature. How does it help to overcome STP limitations? Thanks.

regards
alex

Sounds a lot like Cisco FlexLinks...

I've ceen reading some technical documentation about RTG, but after read the 1st one of the two links below, I'm trying to figure out what is the perspective of the two 4200 located in one end of the RTG... As the two 4200 form a Virtual chassis, how it see the RTG, as a virtual layer 2 interface? it's possible to mix different speeds and media in the same RTG?

 

http://www.juniper.net/techpubs/en_US/junos/topics/example/cfm-ex-series-redundant-trunk-groups.html

http://www.juniper.net/techpubs/en_US/junos/topics/concept/cfm-redundant-trunk-groups-understanding.html

RTG is configured on a physical interface, not a "virtual" interface.  To the switch, (virtual chassis or not) the secondary interface is "down" until the primary link is lost, then it will be "up". 

Hello DN,

 Can you tell me the convergence time of RTG ?

Hi Syq

 

I have made some tests in this kind of topology:

 

              ________  ge-0/0/0 Main    ________________________ 

   ge-0/0/3  |        |-----------------|                        |

Sender Iperf-| EX3200 | ge-0/0/2 Backup |  L2 switch without STP |- Receiver Iperf

             |________|-----------------|________________________|

 

 

With this precise config on the EX switch ( Iperf sender is 10.0.0.4 and receiver is 10.0.0.3) :

set interfaces ge-0/0/0 unit 0 family ethernet-switching vlan members INTERCO
set interfaces ge-0/0/2 unit 0 family ethernet-switching vlan members INTERCO
set interfaces ge-0/0/4 unit 0 family ethernet-switching vlan members INTERCO
set interfaces vlan unit 0 family inet address 10.0.0.1/29
set ethernet-switching-options redundant-trunk-group group TEST interface ge-0/0/0.0 primary
set ethernet-switching-options redundant-trunk-group group TEST interface ge-0/0/2.0
set vlans INTERCO l3-interface vlan.0

 

I tested this with Iperf sending 1000 frames per seconds and counting the number of frames lost each time I disconnected the main interface: ge-0/0/0

 

The results is that I had between 31 ms to 120 ms of frames lost with an average of 76 miliseconds on 7 tries.

 

Theses are the results of the loss during disconnection on Coper interfaces (synchronised at 100mbps) and I believe it depends on the physical type of interfaces as the time is mainly due to detection of loss of signal.

 

During the recovery of main interface I mainly saw no loss.

 

I hope that this will help you

 

 

Would it be correct to say that RTG needs to be configured between the same chassis whether real or virtual and so it may be as well to configure aggregated interfaces instead of RTG so as to use the available bandwidth?

 

Regards

 

Aindriu

Hi,

You would rather use aggregated interfaces instead of RTG between only two equipements.

But in case you have switch A connected to switch B and C you can't have a LAG (agregated interface) with two link from A going to two different switches.

In this case RTG is the solution instead of using STP.

         |          |--------------------------| Switch B |

         | Switch A |                           

         |          |--------------------------| Switch C |

 

I hope this is more clear like that.

 

 

 

Thanks Loup2 that makes sense

 

Aindriu

Your logic is correct.

 

Of course, now JUNOS supports LAG across multiple non-VC switches... not that every implementation would necessarily support it, but those that do can now double their available bandwidth capacity and not lose failover capabilities despite traversing multiple switches.

 

 

I'm glad I read this.

 

I'm still a noob when it comes to Juniper so, thank for your patience! 🙂

 

I've been trying to plan how to successfully integrate spanning-tree to my EX4200 from my Cisco core switches.

But according this, I guess it's not needed.

 

My question is:

 

Because my cisco core switches are sending BPDUs through the rest of the cisco network, what will my EX4200 do when those BPDU's come down the active RTG trunk?  Will it flood them unecessarily throughout the rest of the VC?

 

My cisco is running PVST+ and the native vlan is set to 2 on the trunk in between all switches (including the EX4200).

 

Do I need to turn spanning-tree off on my EX4200? 

 

Any insight is appreciated!

 

Thanks!

hi danno,

 

can you elaborate 'now JUNOS supports LAG across multiple non-VC switches' ? from which version it is implemented?

 

thks

Would you please give a link about "now JUNOS supports LAG across multiple non-VC switches"?
Thank you.