Just realised my Netgear switch does not support LACP or IEEE 802.3ad - does it matter?

[awediohead]

So I bought a Netgear GS116Ev2 about 18 months ago and it stayed in the box until recently as I had to move house and then became far too ill to do anything very much for about a year.

Now I'm feeling a bit better, I've just discovered it doesn't support LACP: The manual states:

The switch does not support IEEE 802.3ad Link Aggregation or Link Aggregation
Control Protocol (LACP) groups but supports manual static LAGs only.

My router running OPNsense does support LACP and so does my Unraid server, so I'm a bit peeved that I bought a switch that doesn't.

My question is: Does this matter? What does it mean I can and can't do?

In OPNsense the options for LAG protocol are LACP, FAILOVER, FEC, LOADBALANCE and ROUNDROBIN so - just out of interest - which of these would work best with the manual static LAG the switch does? I assume either Loadbalance or Roundrobin?

What's going to be much more useful (because our internet is only around 70 Mbps down) is a LAG between the switch and my server which is running Unraid and has a spare 4 x 1000Mbps interface which I'd thought of splitting into two LAGs :-one to pass through to VMs and the other to handle streaming from Plex which could easily be three simultaneous HD streams and lossless audio over the LAN

cheers

 

[drinkingbird]

So I bought a Netgear GS116Ev2 about 18 months ago and it stayed in the box until recently as I had to move house and then became far too ill to do anything very much for about a year.

Now I'm feeling a bit better, I've just discovered it doesn't support LACP: The manual states:

The switch does not support IEEE 802.3ad Link Aggregation or Link Aggregation
Control Protocol (LACP) groups but supports manual static LAGs only.

My router running OPNsense does support LACP and so does my Unraid server, so I'm a bit peeved that I bought a switch that doesn't.

My question is: Does this matter? What does it mean I can and can't do?

In OPNsense the options for LAG protocol are LACP, FAILOVER, FEC, LOADBALANCE and ROUNDROBIN so - just out of interest - which of these would work best with the manual static LAG the switch does? I assume either Loadbalance or Roundrobin?

What's going to be much more useful (because our internet is only around 70 Mbps down) is a LAG between the switch and my server which is running Unraid and has a spare 4 x 1000Mbps interface which I'd thought of splitting into two LAGs :-one to pass through to VMs and the other to handle streaming from Plex which could easily be three simultaneous HD streams and lossless audio over the LAN

cheers

Not a big deal, setting up a static LAG isn't that hard. Static LAG also will not detect a "soft" failure of one link but that is uncommon anyway.

Obviously you can't use LACP mode on the OPNsense. Loadbalance is probably going to be the one that is compatible with the netgear but may have to trial and error it.

Keep in mind you can only exceed 1G if you have multiple flows/transfers going on. A single flow will still cap out at 1G. If both the switch and OPNSense support IP based load balancing it can help more scenarios utilize both links, but I suspect the netgear probably only supports MAC, in which case transferring two files between the same 2 devices will likely still be capped at 1G (though I have seen it work over both links even with just MAC, it depends on the specific scenario and all the devices in the path).

If you truly want >1G you need to look at multigig NICs and switches (2.5, 5, and 10G).

 

[awediohead]

Thanks very much for

Not a big deal, setting up a static LAG isn't that hard. Static LAG also will not detect a "soft" failure of one link but that is uncommon anyway.

Obviously you can't use LACP mode on the OPNsense. Loadbalance is probably going to be the one that is compatible with the netgear but may have to trial and error it.

Keep in mind you can only exceed 1G if you have multiple flows/transfers going on. A single flow will still cap out at 1G. If both the switch and OPNSense support IP based load balancing it can help more scenarios utilize both links, but I suspect the netgear probably only supports MAC, in which case transferring two files between the same 2 devices will likely still be capped at 1G (though I have seen it work over both links even with just MAC, it depends on the specific scenario and all the devices in the path).

If you truly want >1G you need to look at multigig NICs and switches (2.5, 5, and 10G).

Thanks for this - I doubt we currently need >1G for any single flow which is why (aside from wanting to know which of the non-LACP options to choose in OPNsense out of curiosity), what I think might be practically useful is setting up LAGs between the switch and my server - which could in theory be accessed by several family members simultaneously to stream HD 2K and 4K media, access VMs, make back ups, while it's also running the *arrs and other background services, etc, etc. So it's more about just trying to make best use of what we've got and since the server has 4 x 1G ports unused and the switch has at least that many free . . .

It's clear from what you say that there's no "magic" happening with LACP and I can get to much the same place with a static LAG which is easy enough to set up on the Netgear switch. I'm just a bit peeved to note that there's a bunch of 24 port managed switches on ebay that do support LACP for quite a bit less than I paid for the Netgear 16 port switch . . . . grrrrr . . .

 

[drinkingbird]

Thanks very much for

Thanks for this - I doubt we currently need >1G for any single flow which is why (aside from wanting to know which of the non-LACP options to choose in OPNsense out of curiosity), what I think might be practically useful is setting up LAGs between the switch and my server - which could in theory be accessed by several family members simultaneously to stream HD 2K and 4K media, access VMs, make back ups, while it's also running the *arrs and other background services, etc, etc. So it's more about just trying to make best use of what we've got and since the server has 4 x 1G ports unused and the switch has at least that many free . . .

It's clear from what you say that there's no "magic" happening with LACP and I can get to much the same place with a static LAG which is easy enough to set up on the Netgear switch. I'm just a bit peeved to note that there's a bunch of 24 port managed switches on ebay that do support LACP for quite a bit less than I paid for the Netgear 16 port switch . . . . grrrrr . . .

Eh if you have spare ports you can do a 4x to the server and 2x to the OPNSense. May need to check the netgear documentation to see which type they use, probably round robin I'd guess but not sure. I don't do a lot with LAG (do more port channeling on Cisco gear) but I suspect if one side is doing load balancing and the other is doing round robin it will probably coexist fine. Round robin is the most basic and easy one to implement so I suspect that's what netgear is doing.

When running VMs though I prefer to dedicate a NIC to the VM rather than sharing one, so you might want to mix and match, maybe a 2x LAG for the main OS (or the VM that will be heavily used) and 1G normal for the others.

 

[awediohead]

Thanks again - I think I've confirmed that the Netgear Switch is using round robin from a search on the netgear forums. Yes it makes sense to dedicate one port / interface to VMs on Unraid - thanks for the tip.