SG3428x-m2 only 8 link aggregation goups?

Hi,
I was setting up some link aggregations and just realised that the SG3428x-m2 is limited to 8 link aggregation groups.
Unfortunately that's not enough for me as i would be actually needing at least 12.
Is this a hardware constraint or just limited by software for the moment and can be improved / increased?
Having bought a manged switch I would actually expect that it should offer enough link aggregations for setting up all ports (expectation for 28 ports would be 14 lags).
Interestingly enough the product page does not mention any restrictions
Regards
David
- Copy Link
- Subscribe
- Bookmark
- Report Inappropriate Content
most non-enterprise or non-VERY high end L2+/L3 switches are limited to 8 aggregation groups, regardless of port count. this is limited by the switch chip itself, which is not designed by or manufactured by TP-Link and i doubt very much they can do anything to increase it.
On your switch, 8 groups can more than fill the switch (a group can have up to 8 ports)
- Copy Link
- Report Inappropriate Content

Hi @dschmitz
Thanks for posting in our business forum.
The datasheet has written this. (M2 or non M2 are the same in this.)
Consider our layer 3.
Up to 120 LAG groups.
https://www.tp-link.com/en/business-networking/omada-switch-campus/sx6632yf/#specifications
- Copy Link
- Report Inappropriate Content
most non-enterprise or non-VERY high end L2+/L3 switches are limited to 8 aggregation groups, regardless of port count. this is limited by the switch chip itself, which is not designed by or manufactured by TP-Link and i doubt very much they can do anything to increase it.
On your switch, 8 groups can more than fill the switch (a group can have up to 8 ports)
- Copy Link
- Report Inappropriate Content
Hi @GRL
unfortunately it is not enough for me I need to hook up 9 cluster nodes via LACP and trunk two switches (sx3008F and sx3016F) via LACP which makes me need 11 LAGs.
Now i probably take my chances we resolving 3 nodes via linux bonding driver with balance-tlb or balance-alb. But I am still hoping it is just a software limitation here on the switch. Besides strangely enough sx3008F shows in omada that it can be assigned up to 8 LAG IDs.
- Copy Link
- Report Inappropriate Content

Hi @dschmitz
Thanks for posting in our business forum.
The datasheet has written this. (M2 or non M2 are the same in this.)
Consider our layer 3.
Up to 120 LAG groups.
https://www.tp-link.com/en/business-networking/omada-switch-campus/sx6632yf/#specifications
- Copy Link
- Report Inappropriate Content
@Clive_A thanks for your response.
When i bought the switch and had a look at the product page under specifications it did not mention the 8 LAG restriction so the PDF datasheet seems to be more specific.
See https://www.omadanetworks.com/de/business-networking/omada-switch-access-pro/sg3428x-m2/#specifications (Unfortunately I would be needing 9 LAGs of 2.5g 2 LAGs of 10g on the switch so 11 LAGs in total).
Regarding the switch you mentioned this raises the question to me whether I am actually fully understanding LAG/LACP how can this 32 port switch do 120 LAG groups?
I would have assumed 16 (like pairing always two pyhiscal ports) would have been the maximum or does this extend to something else then the physical ports.
Could you shed some light here?
- Copy Link
- Report Inappropriate Content
Hi @dschmitz
Thanks for posting in our business forum.
dschmitz wrote
@Clive_A thanks for your response.
When i bought the switch and had a look at the product page under specifications it did not mention the 8 LAG restriction so the PDF datasheet seems to be more specific.
See https://www.omadanetworks.com/de/business-networking/omada-switch-access-pro/sg3428x-m2/#specifications (Unfortunately I would be needing 9 LAGs of 2.5g 2 LAGs of 10g on the switch so 11 LAGs in total).
Regarding the switch you mentioned this raises the question to me whether I am actually fully understanding LAG/LACP how can this 32 port switch do 120 LAG groups?
I would have assumed 16 (like pairing always two pyhiscal ports) would have been the maximum or does this extend to something else then the physical ports.
Could you shed some light here?
Like GRL wrote, this is merely a chipset limit.
Layer 3 uses using professional and powerful switching chipset and design. Overall, as a core switch, it has to be powerful enough to be a core switch.
The datasheet is not the same as the specs page sometimes. Specs only provide the overall information but not as detailed as the datasheet. Also, refer to the datasheet as a cross-reference.
That 8 limit is written in the datasheet, not in the specs.
Stacking also requires a way more powerful chipset and plans. They just outrun the regular L2+ or L3. That 6000 series is the true core L3 switch we introduced last year.
- Copy Link
- Report Inappropriate Content
- Copy Link
- Report Inappropriate Content
A LAG can have only one port. This is useful if, for example, you only want an LACP lag enabled port (in "active mode" to be able to connect to another switch, which also has a single port "active mode" LACP LAG - nothing else you plug into that would would connect.
They open up the full range of 8 LAGs as you can have any combination of them from 1 to 8 ports so you could have 8x1 port, 1x4 port and 2x2 port , 2x3 port etc etc
For the true L3 switches, they are stackable, in effect one switch can control, effectively "take over" more switches of the same model - they become one MASSIVE single switch in effect,, so it needs to have a larger LAG allowance to cover the potentially hundreds of ports it controls.
- Copy Link
- Report Inappropriate Content

Information
Helpful: 0
Views: 268
Replies: 7
Voters 0
No one has voted for it yet.