Multiple M4300 stacking design and connections

[cw_tec]

Hi!

I'm new to Netgear stacking (coming from Aruba switches) and want to implement a stack of 4x M4300 consisting of:

1x M4300-24XF

2x M4300-24X24F

1x M4300-28G

The 10G switches will carry more load so I thought of a minimum of 2x10G connections between each switch. I added a "backup" connection from the 1st switch to the 3rd one in case of failure of switch 2. Same for the M4300-28G (see attached picture).

Is this advisable or are there better approaches?

Thanks & cheers

Chris

[cw_tec · Accepted solution]

Thanks for your reply.

In the meantime I configured a stack of all 4 units and it works as desired and answered the question myself. The larger switches (24X24F) are connected by 3x10G SFP, the 1st one which acts as master is connected by 2x10G to the next "big" switch, 1x10G to the 3rd switch as fallback line and 1x10G to the 28G. The 28G has also a fallback line to the 3rd switch. I appreciate your concerns about SPOF when it comes to errors. We usually test new firmware releases on redundant devices which are not part of the production environment (have another 14x M4300 of different types within the network). So far not one firmware upgrade went bad, in opposite to some Aruba switches which costed a lot of hairpulling in the past.

I added a drawing of how I connected the switches (green lines = fallback).

Thanks again and have a nice week 🙂

Cheers

Chris

[msi]

> Is this advisable or are there better approaches? If you want to do 1 stack, that would work (just keep in mind that you have to switch the mode of each ports from ethernet to stacking mode + a reboot to switch the port mode active. However it depends ... personally if that is your only stack, I'd perhaps separate into at least 2 stacks. The issue is that if you do updates - while you can do staged updates - if anything goes wrong you have to reboot the complete stack. If there is a bug in your stack, all stack members could be affected at once. I wouldn't do it to be honest. While the M4300 have been solid, bugs occur and having all in just 1 big stack, within 1 single logical unit and fault domain, not what I'd want to work with. While there is a standby stack master elected in a stack, crashes can occur. You might also like to idea of testing a new firmware on 1 switch / stack ahead before you ugprade the more critical or even core switch/stack... if you have 1 single stack there is no easy fire testing possible. It looks like the half-width 10G switch could be your core switch? If you added a second half-width switch besides it and make a stack there, then form a stack with the 2 larger ones and just use port-channel pairs between the stacks? It might depend on your actual need of SFP+ ports, however I tend to use cheaper short (usually red just for marking the importance/difference in role) Cat 6 cables as stack ports so the M4300-12X12F in my case and likely in other situations even smaller 8X8F would sufficient?

[cw_tec · Accepted solution]

Thanks for your reply.

In the meantime I configured a stack of all 4 units and it works as desired and answered the question myself. The larger switches (24X24F) are connected by 3x10G SFP, the 1st one which acts as master is connected by 2x10G to the next "big" switch, 1x10G to the 3rd switch as fallback line and 1x10G to the 28G. The 28G has also a fallback line to the 3rd switch. I appreciate your concerns about SPOF when it comes to errors. We usually test new firmware releases on redundant devices which are not part of the production environment (have another 14x M4300 of different types within the network). So far not one firmware upgrade went bad, in opposite to some Aruba switches which costed a lot of hairpulling in the past.

I added a drawing of how I connected the switches (green lines = fallback).

Thanks again and have a nice week 🙂

Cheers

Chris