English Community

Datacenter NetworkingDatacenter Networking Hardware
All Forum Topics
Options

3 Posts

06-13-2017

US

7 Signins

38 Page Views

  • Posts: 3
  • Registered: ‎06-13-2017
  • Location: US
  • Views: 38
  • Message 1 of 3

Channel-group mode active

2020-08-12, 13:48 PM

Good day,

 

I am working with a customer with Nexus and Catalyst switches that I will be connecting a pair of Lenovo NE1032 switches with.   I am using the Networking Guide for Lenovo HX, October 2019, and they are questioning the channel-group mode active configuration.   In their experience, they have had trouble with it on their Cisco gear.   Can someone give me their opinion on this?   

 

This is from page 19 of the guide:

 

 

The steps to configure the switches are as follows:

1. Issue the following commands on the Lenovo switches are as follows:

G8272-1#( (config)# int Ethernet 1/50-51

G8272-1#( (config-if)# channel-group 80 mode active

G8272-1#( (config-if)#

G8272-1#( (config)#exit

Reply
Options

500 Posts

07-17-2018

US

851 Signins

7462 Page Views

  • Posts: 500
  • Registered: ‎07-17-2018
  • Location: US
  • Views: 7462
  • Message 2 of 3

Re:Channel-group mode active

2020-08-12, 14:24 PM

Hello timster68,

 

Channel-group mode active means that LACP will be enabled on the LAG. You would have trouble with this if the other side is not configured for LACP as well.

 

You do not need to use LACP in order for the LAG to work with VLAG or any other protocol really. If you would rather use a non-LACP lag you can change "active" to "on". This must match on both sides to work properly.

Reply
Options

104 Posts

06-03-2020

US

81 Signins

875 Page Views

  • Posts: 104
  • Registered: ‎06-03-2020
  • Location: US
  • Views: 875
  • Message 3 of 3

Re:Channel-group mode active

2020-08-12, 14:32 PM

Hello,

 

Short answer, most networks today use LACP (mode active), and it is proven in the real world to work between Lenovo and Cisco gear (and is my recommendation for aggregations between Lenovo and Cisco).. Longer answer...

 

This is almost a religious discussion, and each network administrator that has been around for any length of time has had their own experiences with using the various types of aggregation. Each has their pros and cons as noted in the following:

 

Cisco supports the following three modes of aggregation

 

1) Mode "on" - this is a static aggregation - there is no protocol and each side just nails it up and hopes the other side is nailing theirs together also, This is the oldest form of Ethernet aggregation and used to be considered the norm for hooking up different vendors, owing to the fact it was simple and not vendor proprietary. But when using static aggs, if you hook up cables wrong to ports not configured for aggregation, you will almost certainly melt down the network with a loop - Both Lenovo and Cisco support static aggregations.

 

2) Mode "desirable" - AKA PAgP - Port Aggregation Protocol -  in an effort to make aggregations more intelligent (and less prone to loops that happen when using "static" mode "on"), Cisco came up with PAgP. PAgP has the advantage that it tries to ensure the other side of the connection is also running PAgP, and if it is not, tries to protect the network from loops (something that misconfigured static aggs can not do).  Very few vendors support PAgP since it is Cisco proprietary (Lenovo does not support "mode desirable" so it could not be used in this case of Lenovo - to - Cisco connection)

 

3) Mode "active" - AKA LACP - Link Aggregation Control Protocol - IEEE standard that came out to provide a vendor agnostic way to use a protocol to solve the same issues that PAgP was designed to solve. Being a standard all vendors can adopt it and I do not know of any current vendor that does not support it. It's potential down side (IMO) is that the writers of the standard left some things to be decided by the implementers, and this has lead to some issue over the years. Some examples:

3a) The timing of the exchange of LACPDU's can be every 1 second (short timers) or every 30 seconds (long timers). Most vendors, including Lenovo and Cisco default to "Long" timers. But some, like Juniper and Microsoft, default to "Short" timers. A timer mismatch can lead to an unstable LACP aggregation, which is why it is always important to make sure both sides are using the same LACP timers (and since both Cisco and Lenovo default to 30 second timers, there is no need to make any changes when hooking up LACP between the two)

3b) The IEEE specification states that if a port configured for LACP does not receive an LACPDU, it should go into standalone mode, and be treated as an "individual" link. This actually defeated one of the greatest assets of an aggregation protocol, to protect the network from loops when things are misconfigured (and actually made LACP problematic in the early days (a decade ago or more)).  Most vendors, including Lenovo and Cisco eventually decided this was wrong, and most default an LACP port not receiving LACPDUs to a suspended mode, to protect from loops if someone has misconfigured or miscabled ports.

 

Based on my 40+ years in the industry and having lived through the birth and riise of these various forms of aggregation, my personally recommendation today is to use "mode active" (LACP). It is currently the most versatile and most robust. While Cisco maintains PAgP (and even enhances it from time to time), its proprietary nature makes it  (again, in my opinion) not a good choice for most deployments, and rarely between vendors. And static (mode on) should really only be considered as a last resort, usually for attaching to very old switches that may not have LACP.

 

Hope this helps.

 

Thanks, Matt

 

 

 

 

Reply
Forum Home

Community Guidelines

Please review our Guidelines before posting.

Learn More

Check out current deals!

Go Shop
X

Save

X

Delete