1747983 Members
4679 Online
108756 Solutions
New Discussion юеВ

Re: MC-LAG

 
Gandalf123
Occasional Advisor

MC-LAG

Hi to all.

I'm not using Procurve switches from a long time. I've seen that family 3500/5400 and so on, now supports MC-LAG.

As I absolutely need this feature, can I:

1) connect one 2510G to two 3500vl and create a single LACP channel ?

2) connect a server to two 3500vl and create  a single LACP channel ?

I don't understood if the MC-LAG is supported only between the same family (in example, only 3500vl connecting to 3500vl) or if it creates a standard LACP channel

Which model supports MC-LAG, exactly ?

Thank you

5 REPLIES 5
parnassus
Honored Contributor

Re: MC-LAG

Hi, first of all MC-LAG looks like a Juniper acronym...using traditional HP wording (so a broad number of Community users can better understand what you're exactly referring to)...MC-LAG should be then called "Distributed Trunking" or, abbreviated, just DT (where with the "Trunking" part we mean "Port Trunking" as known as "BAGG" which stands for Bridge Aggregation Group or also "LAG" which means Link Aggregation Group, all usual HP/HPE acronyms).

So basically you should find HP/HPE ProCurve (now called Aruba) Switches that support Distributed Trunking against other similar Switches or against some edge devices (like Servers).

So your questions would be reformulated into:

  1. Does one HP ProCurve 2510G support DT against two HP ProCurve 3500vyl switches?
  2. Does one Server (equipped with a multi-port NIC) support DT against two HP ProCurve 3500vyl switches?

I'm sure that (a) part of the answers would be found on the link provided above.


I'm not an HPE Employee
Kudos and Accepted Solution banner
Gandalf123
Occasional Advisor

Re: MC-LAG

Thank you for the prompt response and for the clarifications.

So, If I understood properly, the answer is yes to both of my questions.

Now I only have to find a list of switches that supports DT. Are there any search by features?

parnassus
Honored Contributor

Re: MC-LAG

Mmmm...not of a simple one (like "filter and go") I'm aware of...I mean you would start looking into each relevant HPE/Aruba Switch QuickSpecs here (filtering the QuickSpecs of Switches you think you are going to use)...or, instead, just have a look directly to the HPE ArubaOS-Switch Software Feature Support Matrix (September 2016) (this matrix is specifically related only to - legacy ProVision - ArubaOS based models)...you will then easily find that Bridging Distributed Trunking features like:

  • Distributed Trunking PIM-DM
  • Distributed Trunking PIM-SM
  • Distributed Trunking Primary/Secondary Configuration
  • Distributed Trunking Switch to Server
  • Distributed Trunking Switch to Switch

are supported on these models:

  • HP ProCurve 3500, 5400 zl, 6200, 6600 and 8200 zl
  • Aruba 3810M
  • Aruba 3800
  • Aruba 5400R zl2

provided that you respect the minimum software versions noted per each Switch (and if you are going to update to the latest available software versions then you should be OK with regard of that software requirement).

Statements that are definitely not clear (on first link I provided) are:

"NOTE: DT between different type of switches is not supported.

  • DT is not supported between an HP 5406 switch and an HP 5400R switch.
  • DT is not supported on different platforms that make it generic for the HP 3800 switch and the HP 3500 switch."

What exactly "type" represents is uncertain (the wording IMHO is not clear, at least for me)...probably HP 5400 zl and HP 5400R zl2 are different types of Switch (that's clear) even if they really belong to the same ProVision OS based Switches family (known as legacy HP ProCurve)...this to say you should be careful.


I'm not an HPE Employee
Kudos and Accepted Solution banner
Richard Litchfield
Respected Contributor

Re: MC-LAG

if you are buying new switches, foregt about distributed trunking, and use the much simpler stacking. The stacked switches look like a single entity, and you just set up the aggregation groups across ports on both switches.

Dedicated backplan stacking (with dedicated modules and cables): 2920, 3800, 3810

Frontplane stacking (with standard 10 or 40Gb ports): 5400R

parnassus
Honored Contributor

Re: MC-LAG

Well, purchasing newer Aruba Switches - as per suggested list - means only that Frontplane/Backplane Stacking is just-around-the-corner as a supported feature...and it will be a good thing...but it doesn't exactly mean we're done once we power those Switches on (like a "plug-and-enjoy")...not without dedicating other extra bucks/time/knowledge to correctly deploy the whole stacking idea.

What I'm trying to say is that it's both important to briefly advise the OP about Frontplane/Backplane Stacking existence and to advise the OP about what really means going down the stacking route.

To expand what was cited between parenthesis on the post above:

  • Switches supporting Backplane Stacking: the usage of specific Stacking Modules (one per Switch) and a set of specific Stacking Cables (one or two per Switch depending on - Ring/Bus - stacking topology adopted), without them...forget about any Backplane Stacking deployment on Aruba 2920, 3800 and 3810M, not to speak that stack management will enter in the picture (and must be considered).
  • Switches supporting Frontplane Stacking: the fulfillment of actual VSF deployment guideline's requirements which are various (VSF on Aruba 5400R zl2 shall require a set of prerequisite to be fulfilled before starting, as example: v3 zl2 Modules only, 10G or 40G dedicated links as VSF logical ports, second Management Module functionality/features - if both Switches are equipped with it - will be substantially lost, currently only two Stack nodes admitted, additional MAD Device for managing Split-Brain conditions...just to name a few).

So purchasing a newer Switch doesn't automagically mean that *we're done* (the "enjoy" part)...the apparent (topology) simplicity brought by the Frontplane/Backplane Stacking (so we can forget about Distributed Trunking) has indeed a sort of necessary inner complexity IMHO (in terms of necessary BoM, configurations, requirements, management procedures, etc.) and that complexity should be briefly outlined too...just purchasing a newer Switch is only the very first step of many...doing Stacking is not for everyone (and not for every pocket).


I'm not an HPE Employee
Kudos and Accepted Solution banner