ORLANDO, Fla. -- Cisco rolled out data center network features that automate and optimize massive spine-leaf architectures and introduced a core switch that dwarfs its competition in scale and performance.
The switch, the Nexus 7700,
Cisco is responding to customer demand, no doubt, but it also is seeking to close an opening that competitors might have tried to exploit.
research director, IDC
The Nexus 7700 has 83 Tbps of backplane capacity and it can support 384 40 Gigabit Ethernet ports and 192 100 GbE ports at line rate. The switch, available in July, will hit the market with a new series of F3 line cards, including a 24x40 GbE card and a 12x100 GbE card. These modules will natively support a rich set of features, including multiprotocol label switching, virtual private LAN service, FabricPath and Locator/ID Separation Protocol. Cisco also promises virtual extensible LAN (VXLAN) support.
"It's a behemoth of a switch," said Zeus Kerravala, principal analyst of ZK Research. "It's beefier than anything out there. You could argue that no one needs 83 Tbps today, but this will allow customers to have this product in play for 10 years."
Cisco President of Development and Sales Robert Lloyd said the Nexus 7700 takes up 33% less rack space than other switches with equivalent port density and consumes 60% less power.
"Cisco clearly felt it needed a bigger switch, both for high-end cloud and to counter recent switch introductions from the vendors such as Arista and HP," said Brad Casemore, research director at IDC. "Cisco is responding to customer demand, no doubt, but it also is seeking to close an opening that competitors might have tried to exploit."
Some Nexus 7000 customers who were hoping for more performance headroom in their existing investments might be disappointed by the arrival of the Nexus 7700, Casemore said. However, Cisco has also introduced F3 line cards for the Nexus 7000, including a 12x40 GbE module and a 6x100 GbE module that extend the performance of the existing platform.
"Not everybody will be happy, but Cisco probably has done enough to satisfy most Nexus 7000 customers," Casemore said.
Dynamic Fabric Automation
In addition, Cisco enhanced its Unified Fabric architecture with Dynamic Fabric Automation (DFA), a feature available on the Nexus 6000, Nexus 7000 (with the new F3 line cards) and the Nexus 7700 switches. When an engineer deploys DFA to those switches, the technology becomes available to the rest of the Nexus switches in the data center. When enabled, DFA can automate and optimize a data center network fabric, lowering operational costs for an enterprise, Cisco said. The automation feature also integrates physical and virtual networking.
"Customers continue to tell us that the provisioning process is still pretty manual, and it is a cause of inconsistency and mistakes," said David Yen, senior vice president and general manager of Cisco's data center technology group. DFA is optimized for spine-leaf architectures, but it also can apply to any network design a customer chooses.
DFA can automate the configuration and provisioning of both virtual and physical network elements. A network engineer can define configurations and policies ahead of time in Cisco's Data Center Network Manager (DCNM) software. Afterward, the engineer can plug a top-of-rack Nexus switch into the network, and that switch will automatically provision itself through the Nexus 7700, 7000 or 6000 in the spine. Brocade introduced a similar capability a few years ago with its VCS fabric and VDX switches. DFA also automates fabric configuration for tenants on a virtual network. Engineers can set configuration and policies for the tenant in DCNM. When the tenant adds a new virtual machine (VM), DFA automatically provisions the physical and virtual network resources to support it via Cisco's Nexus 1000V virtual switch.
Read about other Cisco developments
Refreshed campus switching
"Internet of Things" unit unveiled
IT, business goals continue to align
DFA also integrates how the Unified Fabric handles bare-metal and virtualized workloads. DFA enables Layer 3 forwarding on top-of-rack, leaf switches, removing the traditional boundary and complexity between Layer 2 switching and Layer 3 routing.
"We are bringing Layer 3 down to the leaf node; therefore the leaf node represents a distributed gateway across the entire physical network edge," Yen said. "All the chatty Layer 2 bridging protocols are terminated at the leaf, so a VLAN becomes a local concept. Forwarding in DFA will use enhanced forwarding, in which IP addresses are actually used to forward packets regardless of whether communication is within or between traditional Layer 2 subnets."
The DFA-enabled switches handle encapsulation of virtualized workloads on their own, eliminating the need for dedicated gateways. The initial version of DFA will capture virtual traffic with FabricPath-based encapsulation, but later versions of the technology will use VXLAN, Yen said.
DFA provides "automated provisioning, simpler fabric management (physical and virtual) and more efficiency and scalability," IDC's Casemore said. "It connects leaf nodes to every spine switch." It enables Layer 3 forwarding in leaf switches "so that no VM is more than two hops from its destination. It's not SDN, but it addresses problems many enterprises are experiencing. There's no one-size-fits-all in today's networking."
Let us know what you think about the story; email Shamus McGillicuddy, News Director.