Sunday, June 24, 2012

Cisco Intros New UCS Blade, Rack Servers

The new Intel-based systems are aimed at a wide range of workloads, from virtualization and databases to Web 2.0 and big data.

Cisco Systems is expanding its converged Unified Computing Systems portfolio, adding four new Intel-powered servers that are primarily aimed at scale-out environments as well as a new virtual networking card.

The offerings include two new rack servers and a blade server powered by Intel’s new Xeon E5-2400 chips for dual-socket systems, and another blade system running on Xeon E5-4600 processors, for four-socket servers. Intel released the two chip families in May, complementing the Xeon E5-2600 processors that were launched in March.

Cisco officials are leveraging them to expand the reach of their UCS line of converged data center solutions, which the company first rolled out three years ago. The networking giant has since seen its UCS business grow rapidly, now reaching more than 11,000 customers and having an annual run rate of almost $1.3 billion. Cisco also has more than 2,000 channel partners selling the solution.

Cisco CCNA Training, Cisco CCNA Certification

Best CCNA Training and CCNA Certification and more Cisco exams log in to Certkingdom.com


The new systems come three months after Cisco launched UCS 3.0, which included new servers powered by the Xeon E5-2600 chips, expanded fabric capabilities and new integrated networking and virtualization features. The latest offerings complement what Cisco already has done with the solution, according to Todd Brannon, product marketing manager for unified computing at Cisco.

“These new servers round out the UCS portfolio with an even stronger set of products optimized for scale-out and light general-purpose computing as well as a new price/performance [four-socket] category in the mid-range,” Brannon said in a June 18 blog post.

In addition, the new Virtual Interface Card (VIC) will add capabilities to what Brannon said is an important part of the overall story around the UCS.

“One of the key differentiators of UCS is the way in which high-capacity server network access has been aggregated through Cisco Virtual Interface Cards and infused with built-in high-performance virtual networking capabilities,” he wrote. “In ‘pre-UCS’ server system architectures, one of the main design considerations was the type and quantity of physical network adapters required. … In UCS, I/O is effectively taken off the table as a design worry because every server gets its full USRDA [U.S. recommended daily allowance] of networking through the VIC: heaping portions of bandwidth, rich with Fabric Extender technology vitamins that yield hundreds of Ethernet and FC adapters through one physical device. Gone are the days of hemming and hawing over how many mezz card slots your blade has or how many cards you’re going to need to feed that hungry stack of VM’s on your rack server. This simplification changes things for the better because it takes a lot of complication out of the equation.”

Introducing the UCS in 2009 marked a significant change in direction for Cisco. The all-in-one data center solution not only was featuring Cisco’s networking technology, but also Cisco-branded servers, as well as storage capabilities from EMC and NetApp and virtualization from VMware. Cisco became a server vendor, putting it in direct competition with companies like Hewlett-Packard, which had been a strong partner.

It also helped dial up the competition in the converged data center market, sparking others like HP and IBM to grow their offerings. Earlier this month, Dell unveiled its Converged Blade Data Center, which combines Dell’s 12th Generation PowerEdge servers, new EqualLogic storage blade arrays and Force10 MXL blade switching.

For its new servers, Cisco is leveraging the work Intel is doing with processors, Cisco’s Bannon said. The new servers, which are aimed at such tasks as Web 2.0, big data and mainstream IT infrastructure, include the UCS B22 M3 blade and C22 M3 and C24 M3 rack systems, all powered by Intel Xeon E5-2400 processors.

In addition, Cisco is rolling out the B420 M3 blade system, which runs on Intel’s Xeon E5-4600 chips and is aimed at such workloads as virtualization and databases. The system supports 1.5 terabytes of memory and, with the new VIC1225 technology, offers up to 160G bps of I/O bandwidth.

“Intel has brought four socket server options, formerly the province of the mission-critical, ‘EX end of the spectrum, down into the mainstream,” Bannon wrote. “An example of this is our new UCS B420 blade. So if you want four socket core count and performance but don’t necessarily need the comprehensive RAS [reliability, availability, serviceability] features of an EX class system, you now have a price/performance optimized solution for that need.”

Intel introduced the Xeon E5-2400 and E5-4600 chips to ensure that the company is offering processor technology for a wide range of workloads, according to Boyd Davis, vice president and general manager of Intel’s Data Center Infrastructure Group.

“The data center’s a complicated place,” Davis said in an interview with Cisco taped during the Cisco Live conference. “Just delivering application performance isn’t enough. You have to deliver balance, and you really have to fit into the environments that customers are deploying today.”

Tuesday, June 5, 2012

OpenFlow and Cisco OPEN's feedback loop

The northbound SDN interface may limit OpenFlow's role and thwart commoditization

During its SDN conference call this week with Morgan Stanley and other participants, Cisco officials talked often about how a "feedback loop" is vital in its vision of a programmable network. This feedback loop would provide analytical data on users, sessions and applications gathered from the network to define policies and orchestration routines to be programmed back into the network.

The feedback loop is integral to Cisco's SDN architecture - the Open Programmable Environment for Networking, or Cisco OPEN. With a southbound interface, policy, orchestration, operational support and billing support applications tell the network what to do; but a northbound interface takes intelligence from the network to share with those applications so the applications themselves can make intelligent programming decisions based on that network intelligence.

Cisco CCNA Training, Cisco CCNA Certification

Best CCNA Training and CCNA Certification
and more Cisco exams log in to Certkingdom.com


RELATED: Cisco software defined-networking products on tap in June
The northbound interface might be the key to Cisco's SDN strategy, as Brad Casemore points out here. OpenFlow is more of a southbound interface that directs switches from a centralized controller. The Open Networking Foundation, of which Cisco is an active participant, is looking to ruggedize and standardize OpenFlow as much as possible for this task.

RELATED: OpenFlow, Software-Defined Networking and the Enterprise WAN
But Cisco doesn't want its switches to be dumb servants to a one-way southbound SDN interface, be it OpenFlow or any other. Cisco prides itself on building intelligent networks. It invests billions in custom ASICs and software to build up and maintain intelligence in its hardware, and it makes many more billions back from its 70% share in switching, 80% share in enterprise routing and nearly 60% share in service provider routing. This is the value that Cisco refers to when it says, as it did this week in the Morgan Stanley call, that SDNs should harness the entire value of the underlying network.

This is why Cisco is proposing Cisco OPEN, which is intended to open up the network with APIs at layers other than just the data and control plane, which is where the popular OpenFlow protocol is targeted. Cisco may cede the southbound data/control plane interface to OpenFlow; but all other interfaces at all other layers in Cisco OPEN are likely to be Cisco, including the northbound one that collects network analytics and delivers it to the control applications and creates the feedback loop -- or to Cisco, the value of SDN.

Cisco said during the Morgan Stanley call that SDNs and OpenFlow will not commoditize its hardware. Hardware will only be commoditized when value is not created. OpenFlow in and of itself does not create value for Cisco, even though IDC says OpenFlow hardware, software and services will be a $2 billion market by 2016. But OpenFlow within the context of Cisco OPEN does create value for Cisco.

Rather than commanding - controlling -- the SDN army, OpenFlow becomes just another foot soldier with a specific role in the lower ranks of the Cisco OPEN feedback loop.