LAG SCG200 Data Plane (link aggregation / EtherChannel)

  • 1
  • Question
  • Updated 2 years ago
  • Answered

Is it possible to work with LAG, or Etherchannel on the SCG200? I need to set up 2 Data Plane, however, if 1 falls, I'd like to use the other to avoid unavailability.
Photo of Jardel Almeida

Jardel Almeida

  • 50 Posts
  • 4 Reply Likes

Posted 2 years ago

  • 1
Photo of pilot1199


  • 14 Posts
  • 3 Reply Likes
It does not support currently. We have asked them many times before but they are still evaluate that.
Photo of Jardel Almeida

Jardel Almeida

  • 50 Posts
  • 4 Reply Likes
Photo of Albert Pierson

Albert Pierson, Employee

  • 175 Posts
  • 150 Reply Likes
The two 10G Fiber ports on the SCG-200 Data Planes are bridged together.
They do not support Link Aggregation or Etherchannel (which aggregates the two ports together for additional throughput)

They can be conected to two switch ports supporting STP port redundancy, to allow continued operation if one switch port, fiber or Data Plane physical port failure.  The Data Plane will forward BPDU's when feature is enabled: (pg 70 SCG200 CLI guide version 3.6.1 -

ruckus(config)# data-plane [ name ] [name forward-stp

The Data Planes are hosts and will not block the connection, just forward BPDU's.  The Switch ports must support STP and disable one of the ports to prevent loops.

You can also use "Access Core Separation" feature on the SCG-200 Data Planes to increase potential bandwidth and prevent users on the AP network from accessing the northbound core network. This feature creates VLAN's and uses static routes to separate traffic.  The switch ports must be assigned individually to the correct VLAN's so traffic is bridge to only one network at a time.
This system will not allow for redundancy feature as described above.
Details on Access Cores Separation can be found in the SCG-200 Admin Guide pg 59 -

The SCG-200 has two Data Planes.  Each of these are independent processors connected internally to the SCG-200 Controller via a PCI bus.  Each Data Plane has two physical interfaces that can be uses as described above.

For redundancy, if only two 10G Fiber ports will be available in the Data Center - it is better to use both Data Planes then physical redundancy using two fiber ports on one Data Plane.  This doubles the capacity of the system and provides full redundancy as the Access Points Data Tunnels will fail over automatically if one Data Plane stops responding, up to the maximum number of 10K AP's per Data Plane.  Access Core Separation could be enabled on Each Data Plane and still have AP tunnel redundancy.

Photo of Jardel Almeida

Jardel Almeida

  • 50 Posts
  • 4 Reply Likes
Thanks for the clarification, we are studying this possibility of STP

Photo of Dionis

Dionis, AlphaDog

  • 80 Posts
  • 48 Reply Likes
In addition to Albert's note, it is worth noting that if we support LAG on the dataplane ports, you essentially eliminate redundancy.  The whole purpose LAG is to increase bandwidth capacity, thus allowing to support 20Gbps using both 10Gbps ports as a LAG and not solely for the redundancy aspect. 

Being that the SCG 200 is classified and mainly used for very large networks such as carriers and operators, it is essential that oversubscription is not allowed so that true redundancy can be accomplished when using both ports on a single dataplane interface.  As such, this ensures that you always full 10Gbps capacity on each port even if one of the two ports fails (either via switch failure, physical transport failures or otherwise).  The idea is that using a bridge allows us to ensure no more than 10Gbps is allowed through the port so that we have one other port able to support that. 

The controller redundancy mechanism already performs load balancing between the two dataplanes (all four ports, used as bridge and as 2 interfaces).  This means that if you have 10k APs as currently supported by the SCG-200 connected, and use both DPs, then you are essentially allocating 5k on each dataplane.  Should one of the two dataplane fail, the failure would be detected and traffic or tunnels are moved to the other dataplane so that we can still support all 10k tunnels.

Hope this helps.

Photo of Jardel Almeida

Jardel Almeida

  • 50 Posts
  • 4 Reply Likes
OK! Thanks a lot for the explanation Dionis, taking advantage of the forum, I am intending to change the IP of the Dataplane, I would like to know if this change is imperceptible / transparent to the end user, or if he realizes, and is there any impact?

"Get tunnelmgr"
Tunnel Establishment: Enabled
Tunnel Authentication: Enabled