Re: Static and dynamic Ether-channel

prabhu.natarajan · ‎06-14-2013

.

Hi,

am looking for some help in understanding where do we need Static Etherchannel configuration and where do we go for Dynamic Etherchannel configuration (Pagp/LACP).

for example,

| | ------------------------- | | | |

switch A | ------------------------- | switch B | |switch C |

| | --------------------------------------------------- | |

Here, trying to form etherchannel on switch A with four member links in which two links to switch B and two links to Switch C. In this scenario, whether we should go to manuall ethet-channel or dynamic etherchannel (Pgap/LACP).

Please help explain this.

Thanks,

Seeni

glen.grant · ‎06-14-2013

I guess it's matter of opinion, myself between switches I always use negotiated etherchannels that way you know all the parameters are correct . Myself I have always used pagp non-silent) between cisco devices (proprietary) . Have never had a single issue with this , it just works . Others will say use LACP because it is the IEEE standard but it doesn't really matter if it's between cisco devices . Use LACP between cisco and other manufacturers devices.

Caution

You should exercise care when setting the mode to on(manual configuration). All ports configured in the

on mode are bundled in the same group and are forced to have similar characteristics. If the group is misconfigured, packet loss or spanning-tree loops might occur.

Steve Fuller · ‎06-15-2013

I agree with Glen... for the most part

Using PAgP or LACP will ensure the link parameters match before adding links to a LAG, they'll find unidirectional inks, links cross connected between different port-channels etc.

Using static mode means the port-channel will establish quicker, but there's no real checking whether things are configured correctly. For example I've seen a port-channel between a ESXi server and s switch form with one link working at GE and the other at FE. Both switch and server were happy, but we had real strange behavior.

PAgP vs. LACP can comes down to personal preference as they both pretty much do the same thing. I personally prefer LACP for exactly the reason that Glen states, that it's an open standard. This means I can use it switch-to-switch, switch-to-server, switch-to-firewall etc., so I only have to understand and support one port-channel mechanism.

My simple rule is, if I can run dynamic e.g., LACP, then I do, I'll only run static when one of the link partners doesn't support dynamic e.g., VMware ESXi without Nexus 1000V.

Regards

Rolf Fischer · ‎06-15-2013

Glen,

I'm reading with interest that you prefere the non-silent option.

Do you see an advantage in using that over the (default) silent mode combined with UDLD on the member-links?

Thanks.

Best regards

Rolf

glen.grant · ‎06-15-2013

No reallly just going by cisco docs as shown below . It will work either way , I just chose non-silent because it's between cisco devices. Very few non cisco devices use pagp because it cisco proprietary .

If your switch is connected to a partner that is PAgP-capable, you can configure the switch interface for nonsilent operation by using the non-silent keyword. If you do not specify non-silent with the auto or desirable mode, silent mode is assumed.

The silent mode is used when the switch is connected to a device that is not PAgP-capable and seldom, if ever, transmits packets. An example of a silent partner is a file server or a packet analyzer that is not generating traffic. In this case, running PAgP on a physical port connected to a silent partner prevents that switch port from ever becoming operational; however, the silent setting allows PAgP to operate, to attach the interface to a channel group, and to use the interface for transmission.

prabhu.natarajan · ‎06-15-2013

Hi All,

Thanks for sharing your view on this, let me summarize what you got from your replies.

1) Its matter of our choice. either we can go for dynamic or Static config (as per Glen comments)

2) Dynamic (Pagp/LCAP) can be used between switches so that both can generate Pagp/LACP packets (as per Steve)

3) Static can be configured between Switch and Server (non packet generating device) but still Pagp non silent mode help

us in such cases. (as per Glen)

Please correct me if i am wrong.

Thanks,

Seeni

Rolf Fischer · ‎06-16-2013

Hi Seeni,

2) Dynamic (Pagp/LCAP) can be used between switches so that both can generate Pagp/LACP packets (as per Steve)

Dynamic, especially LACP (because IEEE standard), can also be used between switches and server, depending on the capabilities of the server. This is widely used in VMware environments (Steve mentioned ESX).

The proprietary PAgP was developed before a standard existed, same story as ISL and 802.1Q trunking. Some non-Cisco vendors still support it, but the majority supports LACP only.

The non-silent option of PAgP adds the detection of unidirectional links, similar to what UDLD does.

I never understood the part of Cisco documentation that says PAgP silent mode can be used if the partner is not PAgP-capable. In this case, how can negotiation be done successfully? And what is the advantage over the on-mode?

I think this is an incorrect phrase. If I'm wrong and the documentation is correct, it should be possible to interconnect two switches, one in on-mode and the other in desirable-silent mode. When I was studying for the BCMSN exam a couple of years ago, I think I tried all possible combinations and on<>desirable-silent didn't work. At the moment, I don't have access to a lab but I'll give it a try tomorow - maybe my notes need a revision ;-)

Hope that helps

Rolf

Steve Fuller · ‎06-16-2013

Picking up on something that Rolf mentions.

....but the majority supports LACP only.

This is also true for some Cisco devices as well now, for example the Cisco Nexus range. These switches only support LACP so even Nexus-to-Nexus or Nexus-to-Catalyst you would have to use LACP if you wanted to use a dynamic LAG.

Regards