Network Virtualization using Extreme Fabric Connect

Network Virtualization Using Extreme Fabric Connect
© 2019 Extreme Networks, Inc. All rights reserved. 152
Configuration-wise it is similar to VRRP in that both a local IP and a virtual DVR Gateway IP need to be
configured (the latter to be used as default gateway by end-stations). Like VRRP, DVR does not have any
SMLT IST dependencies and thus can also be used on user segments which span multiple SMLT clusters.
Like RSMLT, DVR does not use or need a Hello protocol and can therefore scale to the maximum number of
IP interfaces that the platform supports.
However, DVR works with a concept of DVR domains and within and across these domains DVR announces
IP host routes using the DVR domain and Backbone reserved signalling I-SIDs. All of this functionality could
be useful in campus environments where the users are mobile (e.g., high density of wireless users), but is of
limited use otherwise.
The other important thing to remember is that DVR-enabled L2 VSN/VLANs can only be extended to other
DVR-enabled (controller or leaf) nodes or to FA Proxy switches which are themselves connected into DVR-
enabled (controller or leaf) nodes.
Caution
A DVR-enabled L2 VSN cannot be extended to a non-DVR BEB. Doing so would result in
no IP routed connectivity for hosts located behind the non-DVR BEB.
Load Sharing Over Fabric Connect VSNs
The ability of SPB to load balance traffic along equal cost shortest paths is an important property that goes
hand in hand with its fast rerouting capability to ensure that any link or node failure in the core will not just
heal very rapidly but will only impact a fraction of transit traffic traversing the core.
The SPB Fabric is able to store as many equal cost shortest paths between a source and destination node
as it has BVLANs available. In the Extreme Networks Fabric Connect implementation, currently only two
BVLANs are supported and this section is going to explore how effective an implementation using two
BVLANs really is and how it translates for the different VSN types as well as for IP Multicast running within
those VSNs.
We shall start by examining how IP routed flows can leverage SPB’s equal cost shortest paths with IP
ECMP. The example in Figure 89 represents an SPB Fabric where we have four paths between BEB-1 (or
BEB-2) and BEB-3, which are all of equal cost. Fabric Connect is using two BVLANs, so the fabric itself will
only cater for two shortest paths between a pair of two nodes. This example is equally applicable to L3
VSNs and IP Shortcuts.
Figure 89 L3 ECMP Translation into SPB Equal Cost Shortest Paths
Both BEB-1 and BEB-2, acting as an SMLT cluster (with IP gateway redundancy, VRRP Backup-Master, or
RSMLT-Edge), will announce into IS-IS their point of presence for the same IP net 2. BEB-3 has IP ECMP