Solved

Fabric SPB migration from Cisco core

  • 18 October 2020
  • 15 replies
  • 228 views

Hello,

I was wondering whether someone did migrate from a standard 802.1q tagged core like Cisco to a Fabric SPBM with VSP 7400 series and would like to share their lessons learned.

 

Currently:

Cisco 4500x cores -- X460G2-10GE4 (distribution / edge)

Future:

VSP 7000 -- X460G2-10GE4 (fabric attach)

Options I can think of:

1)

VSP 7400 --- Cisco 4500x

  • not sure how to have 802.1q tagging between VSP 7400 and Cisco 4500x and SPBM at the same time
  • Migrate all X460G2 from Cisco 4500x to VSP7400

2)

VSP 7400 --- x460G2-10GE4 --- Cisco 4500x

  • the x460G2 would be a fabric attached client on the VSP7400 side and a standard 802.1q trunking client on the Cisco 4500x side.
  • migrate all remaining X460G2s to the VSP7400 as fabric attached units

Your thoughts?

Thank you,

 

Klaus

icon

Best answer by StephanH 22 January 2021, 10:10

 

The question regard XOS in a row is answered here:

 

View original

15 replies

Userlevel 6
Badge +1

Klaus,

802.1q tagging is a standard and is working for all brands as far as I know and certainly between Cisco and VSP.

I would go for scenario 1.

The attention points for such migration are:

  • to be sure to control the spanning tree topology and be very careful if PVSTP is in use on the Cisco. The EXOS does support it but no the VSP.
  • check if you use SRP or VRRP for the IP redundancy
    • If you use VRRP, the VSP can join the VRRP instances and by changing the priority you have a smooth transition of the L3 functions

Traditional steps are:

  1. bridge VSP and Cisco core’s
  2. configure all VLANs tagged between them
  3. L2: migrate Edge switches from Cisco to VSP
    1. all the uplinks of switch one at the same time to avoid STP and/or loops issue
  4. Migrate the L3 functions

Steps 3 and 4 can easily be swapped

Regards

Mig

 

Userlevel 7
Badge +1

Hello Klaus,

your option one should work fine. You can have links with tagged vlans from the fabric to other (non fabric) switches. After you establised the connection between Fabric and Cisco, you can move your x460 switch by switch. 

In an third step you can move the routing interfaces from cisco to the fabric.

 

Best regards

Stephan

Hi Mig, Stephan,

 

Thanking you both for your replies.

Cosmetical question:

  • Is there a way to have console messages synchronised and the prompt being shown after the message?

On a Cisco I would configure logging synchronous on line con 0 for instance.

Kind regards,

Klaus

Userlevel 6
Badge +1

Klaus,

 

I’m not sure to fully understand your question but…

To have the on/off the console messages on console or SSH session on:

  • VSP: 
    • configure terminal
    • logging screen
    • no logging screen
  • EXOS
    • enable log display
    • disable log display

Mig

Hi Mig,

 

While I want logging on the console I also want the system to press the return key for me :grinning: like this:

Core-1-Hv:1(config)#1 2020-10-20T21:50:18.464Z Core-1-Hv CP1 - 0x00088524 - 00000000 GlobalRouter SW INFO Boot sequence successful

Core-1-Hv:1(config)#

 

instead of this:frowning2:

 

Core-1-Hv:1(config)#1 2020-10-20T21:50:18.464Z Core-1-Hv CP1 - 0x00088524 - 00000000 GlobalRouter SW INFO Boot sequence successful.

 

Klaus

Userlevel 7
Badge +1

Hello tfsnetman,

 

I think you will find the answere here (last message):

 

https://community.extremenetworks.com/extremeswitching-vsp-ers-232923/voss-console-messages-in-current-ssh-session-7749235

 

The command is  '”logging screen enable'”  hen you use telnet and ssh session.

 

Regards

Stephan

Userlevel 6
Badge +1

Klaus,

We don’t have a solution for this cosmetic issue. A tuned console client can do the job but...

Personally I don’t like the automatic return key. It could happen in the middle of a command line.

I use RoyalTSX as console client and a plug-in based on iTerm can be configured for this.

Be careful anyway.

 

Mig

Hi Mig,

Totally agree with “automatic return” - poor choice of words.

I just meant the console or ssh session to come back to where I left it - so I can see where I was with my typing :grinning: .

I found SecureCRT to be very stable both OS X and Windows.

 

Cheers, Klaus

Userlevel 6
Badge +1

Tfsnetman,

You could always use the TAB or “?” to clear the current line :wink:

Mig

Hello everyone,

 

Cannot open a new topic so I was hoping I could ask the question here.

I am having some issues when connecting switches like this:

VSP7400 - x460G2 - x440G2.

Phones on the x460 and the x440 cannot find their TFTP server for instance.

Software on the VSP: 8.2.6.0
Software on the x460 and X440: 22.7.1.2-patch1-17


Can anyone point me in the right direction?

 

Thank you,

Klaus

Userlevel 7
Badge +1

Hello Klaus,

can your explane your situation in more detail, please?

For example, you write the phones cannot reach your TFTP server.  It would be helpful to know what is working, too.
E.g. if the basic communication works.

Can clients reach each other (in the same VLAN/I-SID and in different VLAN/I-SIDS).

Is the TFTP server pingable.

What else does not work.

How are the XOS switches connected to the VSP? Do you use a fabric? Are the VSP7400, the X460 and the X440 connected in one line or in parallel? ...

Hi Stephan,

Everything works fine when connected like this:

VSP7400 - x450G2

VSP7400 - x440G2

When connecting VSP7400 - x450G2 - x440G2 in one line I am having issues.

Assumption:

The VSP7400 is acting as a FA server, the x450G2 as a FA Proxy and FA Client (for the VSP) and the x440 is a FA client again.

Questions:

  • Do I have to map vlans to ISIDs / NSIs on the x440?
  • What about the uplink on the x450 to the x440? Do I tag all Vlans or do I use LLDP to advertise / request them?
    • Or should I disable LLDP and treat the x440 like any other switch?
  • Do I need version 30.x on the x450 and x440 so I can configure FA Proxy?

Thank you,

Klaus

Userlevel 7
Badge +1

 

The question regard XOS in a row is answered here:

 

Hello,

 

Lesson learned: If you have one VSP cluster member connected to a Cisco with a static Vlan / I-SID configuration, make sure you have that same configuration on both VSP cluster members.

 

Klaus

Userlevel 6
Badge +1

Hi Klaus,

In fact you must have all the vlan/i-sid provisioned on both cluster members even if no ports are assigned to a vlan on one cluster member.

This is due to the fact they announce themselves in the fabric with a virtual bmac and then they both can receive packets for any i-sid announced.

When you configure the virtual ist, you get the following warning:

CAUTION : All the C-VLANs need to have an I-SID associated with it.

Regards

Mig

 

Reply