Technical data

DATA CENTER and CAMPUS NETWORKS DEPLOYMENT GUIDE
Deploying Brocade Networks with Microsoft Lync Server 2010 12 of 52
Core Layer
The core layer consists of high-speed, high-performance, and highly available switches, which connect the
aggregation layers and—in smaller environments—the access layer. In many cases, redundant 10 GbE links connect
the different layers, to provide the required bandwidth. The core layer is also known as the backbone; it is the Layer
3 domain that requires the maximum throughput, non-blocking, high-density, low-latency, and highly available design
architecture. That data center core is the source of packets that are forwarded to external entities, such as the WAN
and campus networks.
The core is one of the most important layers to consider in network design. If the core becomes a bottleneck, then all
attached devices behind it are affected as they try to reach external devices. As network traffic starts to proliferate,
having robust equipment with adequate bandwidth in the core to meet network traffic demand is of the utmost
importance. Typically in this layer QoS, Border Gateway Protocol (BGP), Open Shortest Path First (OSPF), and other
Layer 3 features are deployed.
Ethernet Fabric Access Layer
The access layer is the connection point for servers to “access” network services. The data center LAN typically
requires more resources, and in some cases 10 GbE is required to meet application requirements. Ethernet Fabric is
the key feature in this architecture, as it has collapsed three-tier architecture to two-tier architecture by eliminating
the aggregation layer.
Ethernet fabrics decrease the number of hops to create a flatter, faster architecture that is more efficient, while
remaining highly scalable and resilient. This innovative network advancement takes the most prized qualities of
Ethernet, adds much-needed intelligence, and delivers the services needed for today’s virtualized data center and
changing business requirements.
Since the access layer puts many demands on the network, scalability, high performance, reliability, Power over
Ethernet (PoE), and other advanced features are required. Layer 2 is typically deployed at this level, because it
allows a company to scale and servers and services to communicate more efficiently. The typical features configured
at this layer are Access Control Lists (ACLs), QoS, Class of Service (CoS)/Differentiated Services Code Point (DSCP),
Spanning Tree Protocol (STP), Link Layer Discovery Protocol-Media Endpoint Discovery (LLDP-MED), and PoE.
In some cases, network architects have deployed Ethernet fabrics in the access layer to take advantage of Ethernet
fabric benefits. These include:
• Intelligent decision making
• Reduced network complexity
• Simplified management
• Elasticity
• Improved performance and scalability
To support this, Brocade VDX
®
switches incorporate Brocade VCS
®
Ethernet Fabric technology, which uses TRILL
(Transparent Interconnection of Lots of Links) frames in the data path. Further, VCS technology uses Brocade FSPF
(Fabric Shortest Path First), a well-proven link state routing protocol for Layer 2 networks, in the control plane. Link
state routing at Layer 2 is not “new,” unproven, or risky. The Brocade VCS Ethernet Fabric eliminates having to learn
L2 routing protocols, aka TRILL, multipath links, load balancing, Equal-Cost Multipath (ECMP) configuration, or lossless
Ethernet setup. These are handled automatically in a VCS Ethernet Fabric.
Brocade VCS technology can be updated with Intermediate System-to-Intermediate System (IS-IS) in the control
plane, which provides investment protection. A VCS Ethernet Fabric deployment is incremental and non-disruptive to
existing classic Ethernet environments. You can add it one server rack at a time, if you choose. Fabrics using FSPF
get large, and they are stable, resilient, self-healing, and scalable without reconfiguration of existing network
switches. Such fabrics are flatter, since most rely on core-edge topologies with very low path latency.