Network Virtualization using Extreme Fabric Connect

Network Virtualization Using Extreme Fabric Connect
© 2019 Extreme Networks, Inc. All rights reserved. 11
Reference Architecture
The goal of network virtualization is to decouple the physical infrastructure from the network services used
to interconnect distinct user communities and their applications. Users and devices connected to the
network will only see the virtual network to which they belong and are allowed to communicate only with
other devices in the same virtual network. User communities can be kept separate from one another and
made to access only the applications that they need, increasing the security of the network. Each virtual
network holds the addressing (IP routes, MAC addresses), QoS parameters, and security and access policies
that pertain to that user community, increasing the scalability of the network.
Apart from these obvious benefits, true network virtualization brings about greater benefits in terms of
agility in adapting the network to new applications, new users, and business needs. Much like server
virtualization has brought about a transformation in the way applications are managed and deployed in the
data center, a virtualized network infrastructure fundamentally changes the way networks are managed,
providing the ability to dynamically create, modify, or remove services without affecting other services or
requiring maintenance windows.
Extreme’s Fabric Connect, which offers virtualization capabilities for L2, L3, and multicast, is based on the
Shortest Path Bridging (SPB
1
) protocol, delivering the above-mentioned benefits. It provides a scalable
architecture over a dramatically simplified protocol stack (as compared to MPLS or EVPN), which in turn
results in efficiency gains in network design, operation, and maintenance.
An SPB network uses a single instance of Intermediate System to Intermediate System (IS-IS) routing
protocol whereas MPLS and EVPN require and depend on multiple protocols (OSPF, BGP, LDP, PIM, etc.).
As only one protocol is used in the core all service types benefit from the same fast resiliency without any
protocol dependencies.
Shortest Path Bridging Fabric
Shortest Path Bridging (SPB) is rapidly becoming one of the leading network technologies to deliver an
Ethernet based fabric where all networking services, whether IPv4, IPv6, IP Multicast and/or simply L2
VLANs can be decoupled from the physical infrastructure and virtualized to meet the needs and demands
of typical mid to large enterprises.
SPB was originally defined for carrier Ethernet networks to complement and extend carrier MPLS
backbones. The attributes that made SPB valuable for carrier network providers provide a solid foundation
for virtualized enterprise networks based on a dramatically simplified architecture.
SPB is the combination of three IEEE standards (Table 1) that deliver a new paradigm to the way in which
Ethernet-based networks can operate. SPBM uses Mac-in-Mac encapsulation with the edge device source
and destination IP/MAC addressing encapsulated by the backbone MAC addresses of the nodes on the
edge of the fabric servicing the edge devices. This provides a simple forwarding function because the core
only needs to know the shortest path to the target fabric edge node. The MAC address of the edge node
(switch) is referred to as the Backbone MAC (BMAC), which is used for reachability to other SPBM nodes
using IS-IS as the Interior Gateway Protocol (IGP). A node in the core has a very simple job to do as it only
has to look at the backbone MAC address and forward the packet. The Mac-in-Mac encapsulation
mechanism has no IP addressing required in the core, providing stealth for the network infrastructure.
The IEEE802.1ah Mac-in-Mac encapsulation used by SPBM brings an addressing hierarchy to Ethernet
where the network addressing of end-stations and user devices (whether at L2 with MAC addresses or at
L3 with IPv4 or IPv6 addressing) are always seen as being reachable via a fabric node using the node’s MAC
address (BMAC). From any given source node, the destination BMAC uniquely defines a cut-through
forwarding path without any label swapping (MPLS) or hop-by-hop IP routing.
1
IEEE 802.1aq/IETF RFC 6329