11-16-2011 06:40 AM
I am testing the latest VDX NOS v2.1 in my lab.
Could someone please explain me the advantage of the new feature "virtual IP address"?
What is the benefit?
Where is the feature "management cluster" for a VCS fabric to manage the whole fabric from a single IP?
Is it a supported setup to connect the management LAN port back to a normal switchport for a kind of inband management access?
Thank you for your kind help,
12-14-2011 01:39 AM
We have tested two VDX6720 in a VCS fabric, but both devices have their own configuration each.
Even when we telnet/ssh to the virtual IP on the Mgmt port.
We use NOS v2.1.
Initially Brocade announced for NOS v2.1 that all VDX devices in a VCS fabric should share one common config, like a classical stack, which should be accessed via one single IP address.
But I have not found this feature.
12-15-2011 12:11 PM
All I can find is;
The cluster can be assigned an IP address, known as a Virtual IP (VIP) address. When the VIP is
used to connect to the cluster, successful authentication will log in the user to the principle switch
of the cluster.
So... can all config be done from that VIP or do we still have to connect to indivual switches for some of the config?
12-21-2011 12:19 AM
Have a look on the NOS Update from the PR209 Hercules Product Training in your "my education"
As an extract:
The following Configurations are fully ditributed in the VCS fabric with NOS 2.1
- Virtual IP
VCS with fully distributed configuration is not supported in the initial NOS v2.1.0 release!
12-21-2011 02:48 AM
I checked the M04-Logical_Chassis:
It tells us that:
"Virtual IP Address
* Allows administrators to access a VCS fabric using one single IP address"
But from my understanding I am then connected to the Coordinator VDX in the fabric.
Is it then possible to hop to another rbridgeID/VDX switch in the fabric from this and configure it?
If not, what is the sense of the virtual IP?
02-17-2012 04:21 PM
I played around with this virtual IP as well, and it is a very handy feature. The idea is to have one fixed IP address assigned to the fabric, that stays with the fabric, no matter what switches are part of the fabric.
Let's say your principal goes down for some reason or add another switch which has a lower WWN, you would have to remember a different mgmt-Port IP to access your fabric. The virtual IP, on the other hand, gets automatically assigned to the current principal switch when the VCS is re-formed.
Therefore you can create fixed DNS-Names, that will always get you to the same fabric. It goes without saying, that the ssh host keys obviously will change, because you are connecting to a different unit, but it is still a very handy concept.
I remember having issues with this with NOS 2.1.0, but it is working nicely on 2.1.1. Once your fabric constellation changes, however, it takes a few minutes before the virtual IP is available again. It gets re-assigned to the new principal.
Unfortunately, accessing non-local rbridges is not possible, even when loggin into the VCS via virtual IP. :-(
Hope that helps!