lacp port count - best practice

[p11l]

Hello together,

I visited some Cisco courses and in my last one I heared for the first time that there is a best practice with lacp.

I should be aggregate 1, 2, 4 or 8 ports into one channel.

3, 5,6,7 works not "so good" like the obove one.

Can please someone explain me why it is like it is and where I can find additional information about this?

Thanks!

[paul driver]

Hello @Joseph W. Doherty 
agree of course it will work - if it doesn't then what happens in a 2 link PC and one dies how many links left ?  =  1

[Joseph W. Doherty]

---

@paul driver wrote:

Hello @Joseph W. Doherty 
agree of course it will work - if it doesn't then what happens in a 2 link PC and one dies how many links left ?  =  1

---

Yup.  I know that, but perhaps you need to convince @MHM Cisco World ; )

[MHM Cisco World]

from cisco doc.

""Therefore, you can only achieve perfect load balancing, even with random addresses, if you have two, four, or eight ports in the port channel.""

https://www.cisco.com/c/en/us/support/docs/lan-switching/etherchannel/12023-4.html

 

[Joseph W. Doherty]

---

@MHM Cisco World wrote:

from cisco doc.

""Therefore, you can only achieve perfect load balancing, even with random addresses, if you have two, four, or eight ports in the port channel.""

https://www.cisco.com/c/en/us/support/docs/lan-switching/etherchannel/12023-4.html

---

I don't recall arguing for perfect LB on non power of 2 link member counts.  In fact, I supported that understanding, although with the caveat in the real-world, the theoretical ratios are often not seen for various reasons.

I did, though, mentioned the LB ratio diagram you've also posted, isn't always correct on some of Cisco's later platforms (such as on the sup2T with its PFC4), which implemented a modulo 256, which changes the ratios (for the better) for link members counts not a power of 2.

This reference provides both older and newer bandwidth ratios in its reference.

Also about 12 years ago, Peter and I were discussing a posted question about 3 bit vs. 8 bit hashing for Etherchannel.

[paul driver]

@Joseph W. Doherty 
good thread but now my head hurts i need a lie down!  lol

[Joseph W. Doherty]

"good thread but now my head hurts i need a lie down! lol"

Indeed.

OP's question was great because it leads into the subject of how Etherchannel actually works and its actual effectiveness.

At least I avoided bringing up other Etherchannel considerations, such as standby member links, LACP timing, multi-line card compatibility, etc.  ; )

I'll end with, for a short time I worked within a SP environment using 7609s with nothing but 10g ports often bundled into 8 port Etherchannel.  Coming from Enterprise networking, I wondered, is 80g really needed?  It wasn't, more bandwidth was actually needed because it was insufficient!  Those Etherchannels would max out and start dropping traffic!

[p11l]

Thanks for your help!