[j-nsp] vmember limits in EX series stack

[Naveen Nathan]

I'm attempting to retire a cisco 6509 setup, replacing it with an EX4200
virtual chassis configuration (8 linecards). I've run into a warning
when committing the configuration:

warning: Exceeded vmember threshold limit, it is recommended to
have not more than 32752 vmembers

Our current setup has 200 VLANs defined; and can grow anywhere upto 300
in the future.

Additionally we have around 200 downstream cisco switches; each with
redundant uplinks (so 400 ports). A downstream switch will have an
upper bound of 24 distinct active vlans for the trunk port on the
VC.

Therefore the expected real vmember usage is something like: 200*24 = 4800
or 400*24 = 9600 if considering the redundant uplinks to the downstream
switches.

In this case can the vmember warning be safely ignored? I'm assuming
the warning is due to the worst case; where all 8*48 ports in the stack
are specified to trunk all vlans, so 8*48*200 = 76800 vmembers. However
as stated above the actual amount of vmembers expected (as defined on the
ownstream switches) is far less.

To manually specify the members for each downstream switch trunk port
requires a significant amount of administrative overhead. I would prefer
each trunk port just allow all the vlans.

The documentation is pretty sparse to indicate why this limit is imposed
except espouse caution that the switch process may fail/restart without any
substantial explanation. It would be nice what resources are being tied up.

- Naveen