Big Switch – Cloud Fabric: External (Outside the PoD) Connectivity Options – Part 1

Within the Cloud fabric, both the Edge port capacity (Leaf) & Fabric bandwidth (Spine) can scale out in response to demands. This means that instead of investing upfront in capacity/bandwidth, as was the case with Multi-Tier Fat tree architectures, the network fabric is elastic and can grow/stretch over time. Note that this elasticity is a function of Leaf & spine topology, not the cloud fabric itself. A 2-stage CloS built utilising Layer 3 ECMP/Link State protocols between leaf & spine,  will allow scale out in much the same way.

There are a number of ways to integrate a BCF PoD into an existing network, or to enable external connectivity from the PoD. Following is a simple breakdown followed by a few notes.

BCF Connectivity Out of Fabric

Can be achieved via

  1. Dedicated External Tenant
  2. Dedicated External Segment, within a Tenant.
  3. Dedicated Gateway, within Segments

Note that the External gateway router always connects to a segment within a Tenant. It would never directly connect to the System Router.

The connection can be established, one of the following ways.

External Connectivity 2

Ergo, when connecting to another BCF PoD, Options exist for L2 and L3.

  1. When connecting over L2, the fabric mode will decide how many PoDs can be interconnected.
    • Be mindful that this applies to P-Fabric only (VMWare/ DC underlay Use case), and will not apply to P+V Fabric (OpenStack Use case).
    • When considering “Hub & Spoke” for upto 8 BCF PoD interconnections, Note that the “Hub” Pod in this option will be spineless, made up exclusively of a dedicated pair of Leaf switches (single Leaf Group).
    • There is no need for any other Layer 2/3 protocols.
  2. Layer 2 over Layer 3 – VXLAN, can be achieved for upto 3 PoDs. Only the switches terminating the VTEPs need to be Trident 2+ (Other switches in the fabric can be non Trident 2+).
  3. L3 connectivity can be achieved via BGP or Static Routes

When Connecting a BCF PoD to a Non-BCF Network,

  1. If Layer 2 integration is sought: BCF does not run STP natively. By default, all phyiscal/virtual ports have BPDU guard enabled (Configurable). To integrate at layer 2, BPDU guard will need disabling, and a VLAN will need to be added with membership of all Core L2 interfaces, to enable BPDU pass-through.
  2. for Layer 3 integration, BGP or Static Routing is available. OSPF is expected to be supported in the not too distant future.

The followup post to this one, discusses the Layer 3 integration option in more detail.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s