Cisco launched 40 GbE and 100 GbE capabilities across its Nexus and Catalyst switching lines this week at Cisco Live London, beefing up capacity across campus, data center and service provider networks.
To support the souped up capacity, Cisco also launched a set of network virtualization capabilities that will introduce automated configuration in the WAN, and take on VLAN challenges in the data center and cloud networks.
Cisco is promoting the architecture-wide release as a way for customers to tackle the massive amounts of applications and data that must be moved between the cloud and distributed end users who are either mobile or working in campus or branch office environments.
Massive network capacity for users via 40 and 100 GbE is the first step in this transition, said Shashi Kiran, Cisco senior director of marketing in Cisco's data center solutions division. But “just throwing bandwidth at the problem” isn’t enough, he said. Cisco’s new network virtualization technologies can optimize how engineers provision and manage the new capacity.
If customers actually buy into Cisco's new world order, 10 GbE switches could become basic access switches. Meanwhile, 40 GbE would be used for the data center backbone and 100 GbE for interconnecting data centers and in service provider networks.
There's a lot of “if” in that equation at least in the short term, considering 10 GbE uptake is just beginning to hit full throttle. Convincing the average user why they will need 40 or 100 GbE in the short term will be a challenge for Cisco, said Zeus Kerravala, principal analyst at ZK Research. Nevertheless, Kerravala applauded the release and says Cisco needs to stake ground early.
“Right now the only other vendor that has an affordable 100 gig solution is Brocade, so that's a pretty small competitive landscape,” Kerravala said.
Cisco data center switching upgrade: Nexus 40 GbE and 100 GbE details
Customers who have already invested in the Nexus 7000 line can upgrade to either 40 GbE or 100 GbE using one of two new line cards.
The Nexus 7000 M2-Series 2-port, 100 GbE Module delivers up to 32 high 100 GbE ports in one chassis, while the Nexus 7000 M2-Series 6-port, 40 GbE Module provides 96 non-blocking 40 GbE ports. Cisco also launched the Nexus 3064-X, a top-of-rack switch that supports 1, 10 and 40 GbE connectivity.
University of Twente ICT engineer Jeroen van Ingen, who spoke at Cisco Live London this week, will be a Cisco 40 GbE early adopter.
“At the moment we are seeing 10 gig links slowly filling up with higher and higher peaks of traffic. We are seeing peaks of 8 gigs and 9 gigs,” said van Ingen. “The natural next step is to go for 40 gig.”
Cisco pushes 40 GbE into the campus
Cisco has placed much of its marketing focus on Nexus in recent years, but its largest install base is still Catalyst 6500 users. So at Cisco Live London, the company extended its 40 GbE capability to the 6500 with the Catalyst 6900 Series card. This release comes less than a year after Cisco unveiled the Supervisor 2T terabit card that tripled the throughput of the Catalyst switches. Now the 6900 increases that capacity to 4 terabits.
One network engineer at Cisco Live London, who asked for anonymity, said the extended capacity is enticing, but would cause staggering oversubscription problems.
Addressing that criticism, Cisco vice president of data center switching product management, Ram Velaga said there may be “slight” overprovisioning but that it wouldn't be a major problem for users.
“We have figured out how to double the capacity again with no change to the hardware,” said Velaga. Running the slots at 160 enables two ports at 100 gigs. “Customers are able to deal with [this kind of] oversubscription,” he said.
More importantly, he noted, the idea of the Catalyst 6500 upgrade was to ensure further “investment protection” for those who don't want to do a rip and replace. However, users looking for total “non-blocking performance” could turn to the Nexus 7000.
The campus release also includes Catalyst 4500-X, a 10 GbE 1-RU form factor top-of-rack switch that complements the 6500 switching – a hole in the Catalyst line that many users have noted over the years. Because of the small form factor, users can now extend 10 GbE deployments to the top of rack without the need for large boxes and a huge footprint, said Velaga.
Network virtualization changing the data center and the WAN
Among the network virtualization strategies announced at Cisco Live London is the Easy Virtual Network (EVN), a WAN segmentation technology that runs on the Catalyst 6500, Catalyst 4500 and the ASR1000 platforms.
The technology – an evolution of VRF-lite that is also compatible with MPLS – lets users automatically spin up separate logical networks on a shared network. The idea is to be able to scale the network in a simple way. Cisco says these configurations can be made with as little as one command line.
VRF could be seen as a form of software-defined networking with “OpenFlow connotations,” said Kerravala. OpenFlow also lets engineers spin up virtual instances of network resources. However, Cisco's WAN segmentation technology is proprietary.
On the data center side, Cisco also added VXLAN capability to its Cisco Nexus 1000v virtual switch. VXLAN addresses the challenge of not having enough VLANs in the data center, allowing engineers to scale up thousands of VLANs necessary for logical separation in a multi-tenant cloud network. The release also includes the Nexus 1010-X, an upgrade to the 1010 appliance that manages services, such as compliance implementation, in virtualization infrastructure.
Finally, the release is accompanied by a series of software changes that also aim to enable automation and programmability of the network. The most notable is an upgrade to the Nexus OS that will enable engineers to automate data center workflows. The Nexus OS upgrade will also enable Python scripting to customize network behavior. The capabilities are already available on the Nexus 3000 but will be added to the 2000 and 5000 series later this year.
“We've introduced the ability to run Python scripts on the device itself,” said Velaga. Giving a use case example, he said, if there are four ports and one is dropping packets, the engineer could easily shut the fourth port down. Effectively, Velaga said, this is a form of Cisco embracing software-defined networking.