Users Guide

Rack or VLT fabrics

When two leaf switches are discovered on specified VLTi ports, a VLT is automatically created between the two switches to

form a network fabric called the VLT fabric. This VLT fabric is automatically assigned with a fabric ID, a universally unique

identifier (UUID).

In a single rack deployment, the network fabric and the VLT fabric represent the same set of switches. In a multi rack

deployment, each rack has a VLT fabric, and all the VLT fabrics and the spine switches together form the network fabric.

Default fabric settings

SFS automatically builds the network fabric using industry-standard Layer 2 and Layer 3 protocols.

Reserved VLANs

To build fabric, SFS reserves VLANs 4000 to 4094 for internal use. You are not allowed to use these VLANs for general use.

● VLAN 4000 — SFS control VLAN SFS automatically configures VLAN 4000 on all switches that are discovered in the

fabric, and uses it for all fabric operations internally. When a leaf or spine is switch is discovered, the ICL or ISL ports are

automatically added as tagged members.

● VLAN 4001 to 4079 — leaf and spine connections SFS automatically sets up the leaf and spine network configuration

using eBGP as the underlay routing protocol. SFS uses the reserved VLAN range (4001 to 4079) with automatic IP

addressing to set up the peer connections. When a spine switch is connected to the fabric, an ISL is created between the

leaf and spine switch. Each ISL link uses a reserved VLAN and the ISL ports that are configured to be the untagged members

of this VLAN. IP addresses from the reserved range are used for this VLAN, and an eBGP session is started on the VLAN IP

interface.

● VLAN 4080 — global untagged VxLAN VLAN SFS automatically sets up VXLAN overlay networks with EVPN to extend

networks between racks in a multi rack deployment. SmartFabric OS10 requires an untagged VLAN on leaf switches for

VXLAN traffic handling when using VLT. VLAN 4080 with automatic IP addresses from the reserved range is used for leaf-

to-leaf interconnect (ICL) links.

● VLAN 4090 — iBGP peering between leaf switches SFS automatically sets up iBGP peering between a pair of leaf

switches directly connected over ICL links. VLAN 4090 with automatic IP addresses from the reserved range is used for

enabling iBGP sessions between the VLT peer switches.

● VLAN 4094 — VLT control VLAN SFS automatically creates VLAN 4094 on all leaf switches. VLAN 4094 is used for all

VLT control traffic between two VLT peer switches. VLAN 4094 is only added on the VLT interconnect links (ICL ports) on

leaf switches.

● VLAN 4089 — OS10 internal use In SmartFabric mode, VLAN 4089 is the default VLAN and is reserved for OS10 internal

use.

Default client management network

SFS automatically sets up an overlay network that is called a client management network. When a device is automatically

onboarded on to the network fabric, the device uses the VLAN mapped to this overlay network. This network is a native VLAN

unless there is a policy specifying a different native VLAN. VLAN 4091 is used as the default client management VLAN for this

VXLAN network.

NOTE: The embedded SFS user interface allows you to change this VLAN to a specified VLAN.

Default client control traffic network

SFS sets up a second overlay network that is called client control network specifically for VxRail integrated solutions. When a

VxRail node is discovered, it is automatically added as a tagged member of this network. SFS also enables the mDNS Avahi

service on this network for master advertisement and fabric discovery by integrated solutions. The SFS Master virtual IP for

VXLAN network is advertised. The VIP address is fde1:53ba:e9a0:cccc:0:5eff:fe00:1100 is fixed and not user

configurable.

VLAN 3939 is used as the default client control VLAN for this VxLAN network. Although you can change the VLAN associated

with this, it is not recommended to change it for VxRail integrated solution deployments.

SmartFabric Services