In a TurboIron 24x is a static trunk load balanced the same way as an active link-aggregation?

  • 1
  • Question
  • Updated 4 months ago
I have two Brocade TurboIron Ti24x running 8.00.1b.  Between the switches I have ports 18 and 19 configured as a trunk.  If you show the config you can't see it, but I'm pretty sure a few years ago it was configured using this command:

(config)#interface ethernet 18 ethernet 19
Trunks will be created in the next trunk deploy
(config)#trunk deploy

Now we added a new storage array and this array uses 802.3ad LACP link-aggregation, two ports to one switch, two ports to the other switch.  To configure this, we have the storage array set to passive and the switch active.  The configuration on the switch looks like this:
nterface ethernet 21
 port-name Tegile Controller A Port 1
 no spanning-tree
 link-aggregate configure timeout short
 link-aggregate configure key 21001
 link-aggregate active
!
interface ethernet 22
 port-name Tegile Controller A Port 2
 no spanning-tree
 link-aggregate configure key 21001
 link-aggregate configure timeout short
 link-aggregate active
!

Now my question to the community, is there anything different between manually creating link-aggregate vs a trunk?

Should I CHANGE the links between the switches to link-aggregate?
Photo of kjstech

kjstech

  • 3 Posts
  • 0 Reply Likes

Posted 4 months ago

  • 1
Photo of Jon Maiman

Jon Maiman

  • 5 Posts
  • 2 Reply Likes
In the version of code you're currently running Trunk command is used to create static trunks.   Link-aggregate commands under the interfaces is used to create dynamic LACP trunks.     For your use case you want the link-aggregate commands.

--Jon
(Edited)
Photo of kjstech

kjstech

  • 3 Posts
  • 0 Reply Likes
So on the static trunk between the switches, i’ll never max it out at 20gbps unless I tear it down and reconfigure it as a link-aggregate?

I do see bandwidth on both ports, however when I fail that storage controller to B, most I get is like 1.4GBps (14,000mbps). If controller A has the storage pool I can max it out at 2.1GBps since the traffic isn’t traversing the static trunk.
Photo of Jon Maiman

Jon Maiman

  • 5 Posts
  • 2 Reply Likes
Both sides of a trunk (also known as a Link Aggreggate or LAG) need to be configured the same way.   E.g. both the Tegile and the TurboIron should be configured for a static trunk or for a Dynamic LACP trunk.   So if the Tegile is configured for LACP in passive mode, you should reconfigure the TurboIron for LACP in active mode.

Utilization of the member links in a trunk is a different aspect.   A hashing function is used to distribute the traffic across the member links in the trunk.   The more flows you have, statistically the more likely it is  the traffic will be evenly distributed across the member links.   In practice, regardless of it is a Dynamic LACP Trunk or a Static Trunk, you won't max. out the links nor will you get a full 20G.

Finally from your comments about Controller A and Controller B above, is this truly a LACP setup on the Tegile or is it active/passive backup between the controllers?

--Jon
Photo of kjstech

kjstech

  • 3 Posts
  • 0 Reply Likes
Ok the Tegile controller A is LACP passive to TurboIron switch 1, ports 21 and 22, setup as link aggregation (in active mode).

The Tegile controller B is LACP passive to TurboIron switch 2, ports 21 and 22 as link aggregation (active).

So between the switches and the storage arrays, its configured up and operational.

Now, between TurboIron switch 1 and TurboIron switch 2, its configured as a static trunk on both ends.  Because both ends are configured the same, the link is up.

I was just curious if there was any fundamental difference between creating a static trunk vs using the link aggregation commands.  Obviously the link aggregation commands are required for the Tegile array since it speaks LACP.  However between two switches, it sounds like to me you could configure it EITHER WAY, as long as both ends are configured the same.

There's no bandwidth difference between either way right?  As long as both ends are the same, either solution accomplishes the same result correct?