Problem with link aggregation + sshd

[Nikolay Denev]

On Aug 29, 2012, at 1:18 PM, Harald Schmalzbauer <h.schmalzbauer at omnilan.de> wrote:

> schrieb Pete French am 29.08.2012 11:38 (localtime):
>>> Link aggregation can never work with two separate switches! LACP and
>>> static trunking require both sides to bundle the same trunk. which is
>>> impossible for two separate switches.
>> These switches had a port where you could connect them together and
>> then configure each to know about the other switch, and to do LACP
>> across the pair of them. Or at least thats what it looked like it
>> was capable of doing, and it appeared to be doing LACP when configured
>> that way and connected to Windows machines, just not FreeBSD ones. But I'm
> 
> What you desciribe is well known as „stacking“ (not to mix with „virtual
> stacking“) and sorry that I haven't made clear that in such a case LACP
> (also static trunking of course) works well and is a fantastic way to
> gain redundancy.
> When you create a physical switch stack, the individual switches are no
> separate switches anymore, but act like one big switch.
> With the advantage, that in case of a failure, and a trunk configured
> over two different units of the stack, the link remains active.
> But like mentioned, these switches are then not considered to be
> separate („virtual stacking“ only combine them in management regards,
> _not_ physically, so be carefull when you look for switches with
> „stacking“ capabilities!).
> The disadvantage of the real hardware stackable switch is the price. The
> cheapest way I've found is two DGS-3120 (~700$ each plus 200$ stacking
> cable). Ciscos and Junipers and the bigger HPs are all much above afaik.
> 
> -Harry
> 

Not always. For example Extreme Networks's MLAG allows link aggregation between two switches, that
are not stacked. You just have to create a special vlan between them and configure them for MLAG.
But of course this is proprietary protocol.