Hi, James,
On Tue, Nov 26, 2013 at 10:13:40AM +0800, James Andrewartha wrote:
> On 26/11/13 06:13, Geoff Smith wrote:
> > I have used both methods and there are advantages and disadvantages to
> > each depending on the details of your situation. If all things are equal I
> > would recommend using LACP. The primary reason being simplicity of ongoing
> > maintenance/management. I had a situation where I needed to add a few more
> > VLANS to my network and realised that this would change the hash value of
> > my MSTP. I had to create the VLANS on every switch in the MSTP system to
> > make it re-converge correctly (even though the VLANS were not required in
> > all locations). This all turned out OK in the end but in the process I
> > managed to badly break the entire system as the MSTP trees fragmented
> > during the process of adding the new VLANS. Luckily I was doing this
> > during a maintenance window!!! By contrast adding a new VLAN on a LACP
> > system is relatively painless ( just make sure that you egress the new
> > VLAN/S on both the LAG port and the underlying physical ports).
>
> Ahh, I wasn't aware that adding a VLAN would change the MST hash and
> cause re-convergence. Given this I might just have my core switches
> (which have every VLAN) in the same MSTP region, and will leave the
> access switches using LAGs and RST.
Moving a VLAN from one instance to another changes the hash. All possible
VLANs are allocated to one of the instances -- if you have the default
configuration, they are all in instance 0.
You could e.g. create two MST instances (SIDs), allocate VLANs 1-2047
to instance 1 and VLANs 2048-4093 to instance 2. When creating a new
VLAN you chose the ID from the range of instance 1 or instance 2 for
per-VLAN load balancing.
The allocation of VLANs to MST instances is independent of the VLANs
actually existing on the switches.
Anyway, I'd suggest using LAGs instead. Easier operation, perf-flow as
opposed to per-VLAN load balancing, automatic instead of manual load
balancing. If a single link inside an LAG fails, only two switches are
directly affected. If a single link in the spanning tree changes, other
switches might need to run the STP algorithm as well.
When using LAGs, consider "set lacp singleportlag" _and_ disabling LACP
on all ports that are not part of a LAG ("set port lacp port PORTSTRING
disable"). (If singleportlag and LACP is enabled on an inter switch link,
this link will try to from a LAG. This may happen with LACP capable servers
connected to the switch as well.)
Best regards,
Erik
--
Dipl.-Inform. Erik Auerswald http://www.fg-networking.de/
[email protected] T:+49-631-4149988-0 M:+49-176-64228513
Gesellschaft für Fundamental Generic Networking mbH
Geschäftsführung: Volker Bauer, Jörg Mayer
Gerichtsstand: Amtsgericht Kaiserslautern - HRB: 3630
---
To unsubscribe from enterasys, send email to [email protected] with the body:
unsubscribe enterasys [email protected]
