On Monday Cisco’s carefully curated press program at Cisco Live Orlando focused on the refresh of the company’s campus switching portfolio, most notably with the new Catalyst 6800 series of chassis switches whose supervisor modules and line cards are backward and forward-compilable with the Catalyst 6500.
Meanwhile, Cisco Live attendees got a special treat during a keynote presentation by David Yen, vice president and general manager of Cisco’s data center technology group. Yen shared the stage with the a very tall member of the new Nexus 7700 series of data center core switches. This new family of switches have up to 83 Tbps of total switch capacity, with 1.3 Tbps of throughput on each slot. A fully populated top-of-the-line Nexus 7700 can support 384×40 Gigabit Ethernet (GbE) or 192×100 GbE ports. The switch had attendees swooning.
Holy cow!! The new Nexus 7700 looks awesome & the throughput made my jaw drop! In addition F3 line cards! Literally blowing my mind #clus
— nateb72 (@nateb72) June 24, 2013
Cisco Nexus 7700. 42tb of switching capacity? Holy hell that’s a lot of capacity. 1.3tb/slot #clus
— Ian Underwood (@iunderwood) June 24, 2013
When details of this monster switch leaked out over Twitter, I asked the Cisco PR team about it. They were tight-lipped. The official announcement of this switch is scheduled to take place during a Wednesday press conference, and Cisco isn’t prepared to offer details to the general public yet. Some content about the switch appeared on Cisco’s web site recently, but it was scrubbed when some network engineers stumbled upon it.
Given that Insieme Networks executives are scheduled to participate in the Wednesday press conference, I expect the news will be much bigger than just the Nexus 7700. My sources have told me that Insieme has been developing a massive fabric controller that can orchestrate the entire data center, not just the network. Rumors have persisted that the stealth Cisco spin-in is deeply involved in software-defined networking, too. But I’ve never been able to confirm that rumor.
I’ve heard Cisco executives boasting that this year’s Cisco Live will feature one of the biggest collections of major new product announcements in years. The Catalyst 6800 was a good start and the Nexus 7700 was impressive, but I think there’s more to come.
John Curran, the voluble president and CEO of the American Registry for Internet Numbers, didn’t waste any time exhorting attendees at last week’s North American IPv6 Summit in Denver to break the logjam delaying the widespread implementation of the next-generation protocol.
“Why are we doing this?” he asked. “What is the one event” that will spark the momentum needed to fuel IPv6’s adoption?
Curran said the energy sparked by the realization that IPv4 addresses would soon disappear had sputtered over the past year as enterprises and ISPs found other ways to manage device identification and addressing.
“ISPs say customers aren’t asking for [IPv6], and you can’t expect ISPs to deploy when customers aren’t asking for it,” he said. And why aren’t customers demanding their providers switch to IPv6?
“Because they believe they are already connected to the Internet. We must disabuse them of that notion.”
To Curran, IPv4 is not the Internet, users’ claims notwithstanding.
“You are connected to a subset of the Internet,” he said of those users that believe they’re fully Web-enabled.
“We have to begin to tell customers they are not on the Internet; they are paying to be on, but they need to be told they are not on the Internet unless they are on IPv6,” Curran said.
To be sure, Curran and ARIN have a vested interest in encouraging enterprises and ISPs to adopt IPv6. North American IPv4 addresses will be exhausted by 2015, and with millions of user devices and other Internet-aware gadgets slated to come on-stream in the next few years, IPv6 is the only alternative. But IPv6 adoption, at least in the United States, has been glacial. While a little more than a third of U.S. government websites are IPv6 enabled, only 3.7% of industry websites and 5.7% of educational websites are similarly supported, according to stats shared at the Summit.
And that doesn’t say anything about how few enterprises’ internal networks natively support the IPv6 protocol.
For better or worse, many ISP and enterprise executives remain reluctant to invest the time, money and resources necessary to migrate to IPv6. The protocol’s proponents, and there are many, understand this. But they also understand how critical it is for companies and carriers to embrace IPv6.
Not because it can handle as many addresses as there are grains of sand, as the saying goes. But because it will also usher in new services and new capabilities that U.S. businesses and ISPs will need to remain competitive in the years and decades to come.
Over 500 attendees join the 2013 North American IPv6 Summit today — the highest in the conference’s history. The Rocky Mountain IPv6 Task Force (RMv6TF) in conjunction with the Regional North America IPv6 Task Forces are throwing a three-day long conference, April 17-19, dedicated to the next generation of the Internet: IPv6. First-day attendees get a day-long tutorial and hands-on experience for a chance to tinker with a real live IPv6 environment. The following days feature educational talks from IPv6 super-stars, like Google’s Latif Ladid and ARIN president John Curran. These experts and many others are experienced in the ways of the new Web. Attendees are joining for lots of reasons, but mainly, not to get left behind in the ever-evolving IT industry. With the depleted pool of IPv4 addresses and IP-enabled devices entering the market, the time to know about IPv6 has never been more crucial.
Cisco began “limited restructuring” this week by laying off approximately 500 employees globally, including two of the four -person team manning the Cisco/EMC alliance, a source tells us.
Ties between Cisco and EMC had been visibly strained since VMware — a company owned by EMC — acquired network virtualization provider, Nicira Networks Inc. last summer. While Cisco had historically been an ally with VMware and EMC, EMC/VMware’s acquisition of Nicira’s network virtualization left a bad taste in Cisco’s mouth.
Another indication of the souring relationship between Cisco and EMC this week is a rumor that EMC will start selling its own white labeled servers, putting it into another field of competition with Cisco — a rumor unconfirmed by Cisco.
While Cisco’s actions this week may point to a souring Cisco/EMC relationship, it could also indicate Cisco’s ongoing shift toward focusing on software and cloud initiatives. As demand increases for software-defined networking (SDN) and network virtualization solutions, Cisco could feel pressure on its traditionally robust hardware profit margins — an industry trend that is apparently on the minds of Cisco execs as the company works to realign operations.
“We routinely review our business to determine where we need to align investments based on growth opportunities…These actions are subject to local legal requirements, including consultation, where required,” said Cisco spokesperson Robyn Jenkins Blum, Cisco Corporate Public Relations in an email response.
Alcatel-Lucent recently put up its patent portfolio as collateral for a credit line from Goldman-Sachs and Credit Suisse, and this arrangement has the French government nervous. The government may urge the company to sell off business units instead, including its enterprise division.
According to a report in the French business publication Les Echos, the government is discouraging Alcatel-Lucent (ALU) from using the 1.6 billion euro credit line. The government fears that ALU could default on its loans and let its patents fall into the hands of foreign banks. The credit line would stabilize ALU while it tries to reduce costs following consecutive quarters of heavy losses. The French government prefers that ALU find alternatives to putting its patents up as collateral. My French is rusty, but my reading of Les Echos reveals that the government is pushing ALU to sell off a valuable business unit to raise cash, including the enterprise business or its submarine communications business. The government is also exploring a patent consortium, which would allow ALU to share its patent portfolio with other companies and derive revenue from them.
ALU has put its enterprise business up for sale a couple times in recent years, so this news is not great shock. But some customers must be frustrated by the continued instability of the company. ALU has a lot of debt, negative cash flow and dwindling cash reserves. Seeking Alpha says the credit line is unlikely to turn around the company’s fortunes.
Everyone has heard of bringing your own device to the workplace, or the BYOD trend. While many networking products have been ramped up over the past several year to accommodate more devices entering the enterprise network, the next-generation trend has entered the arena and attention must be paid.
Bring-your-own-application, or the “BYOA” trend is demanding a wireless LAN prepared to handle not just new devices, but the slew of new applications being brought into the workplace. Aruba Networks has recently announced a new wireless LAN platform with Aruba’s AppRF technology — a series of three mobility controllers, the 7210, 7220 and 7240 — aimed at addressing the onslaught of mobile applications and devices coming onto the enterprise network.
“Our customers were coming to us and saying, ‘we have a problem not being able to control the applications running on our employee’s mobile devices on our network — We don’t want to block them, we just want to know what the biggest bandwidth consumers of Wi-Fi are,’” said Ozer Dondurmacioglu, Aruba’s director of product and solutions during a briefing.
And while IT has been able to manage WAN bandwidth consumption, Wi-Fi bandwidth has been a different story.
“For Wi-Fi, everything changes as devices move around, and we wanted to help with capabilities to show [IT] what is going on in the air,” he said.
Enterprises will be able to eliminate desktop phones, IP PBX support, expensive video and audio equipment and dedicated videoconferencing systems thanks to the new platform, which can guarantee high quality and performance for popular voice, video and Unified Communications (UC) applications, like Microsoft Lync, Dondurmacioglu said.
“The new normal is users love their mobile devices — they wouldn’t want to use anything else, even if IT begs them,” Dondurmacioglu said. “This year is going to be a battle between how to manage mobile apps over the air, and how to manage what is most important to an organization is protected given the limited Wi-Fi bandwidth.”
In Brocade’s latest quarterly earnings call with Wall Street analysts, executives revealed that its VCS data center fabric and VDX switches are winning new customers and expanding their footprint in existing accounts. Also, VCS fabric customers are telling Brocade that the technology will help them migrate to software defined networking and network virtualization in data centers.
According to a SeekingAlpha.com transcript of the earnings call, CTO David Stevens said sales of the VDX switching line are accelerating and expanding. In the first year the technology was on the market, Brocade saw mostly pilot projects, but “now we’re seeing a fair number of those accounts scale out into broad production use of the technology. In fact, some of the customers [are] hitting the limits of” the original VDX architecture.
Brocade announced the VDX 8770 chassis switch this year to increase the scale of the VCS fabric. The company now has 800 VDX customers.
“Over time, we’re going to see more scale-out production use of the technology, both… within [the] installed base where we sold the product to date but also as we gain new name accounts going forward,” Stevens said.
In its final quarter for fiscal 2012, Brocade reported $578 million in revenue, a 5% bump year-over-year. It was a record quarter for the company, driven mostly by a robust sales in storage area network (SAN) sales. Its IP networking business declined by 3%, pushed down by routing. Switching actually grew by 5%.
During the earnings call one financial analyst, Mark Sue of RBC Capital Markets, pushed Brocade’s executives on the idea that it should focus its Ethernet business in the data center, saying “the business might benefit from some focus… Is there some thought of driving that deeper into the data center and perhaps less in the campus and less in the enterprise just because the market doesn’t seem to be growing that margin? It is very crowded.”
Jason Nolet, VP of Data Center Networking Group, said Brocade has invested substantially in its VCS fabric and its VDX switches. Investments in campus networking aren’t taking away from that data center focus, he added. Stevens, the outgoing CTO, added that investments in campus networking are relatively small compared to the investments the company has made in developing VCS and service provider networking.
Brocade started refreshing its campus networking products a year ago with the ICX 6610 series. Next year it will release HyperEdge, a campus LAN management technology that establishes a single management IP address where admins can make changes to an entire network through a single CLI session.
Stevens added that customers are starting to engage with Brocade about the need for software defined networking technology, especially for implementing network virtualization.
“I think it’s starting to gain a lot of interest,” he said. “When you think about adding another layer to the network with network virtualization, you’re going to add logical networks through tunnel technology. You’re actually adding to the overall administrative burden of that environment, because the physical infrastructure doesn’t go away. It still needs to be scaled, maintained and managed to upgrade, et cetera.”
Customers are telling Brocade that the VCS fabric’s ability to “simplify and reduce the operational overhead of that underlying transport as a result of the very high level of automation and efficiency that we’ve built into the fabric” allows them to focus more on how they’re going to deploy and run network virtualization, Stevens said.
“It also prevents them from just doubling up their operational overhead as a result of having adding that additional virtualization layer to the network environment,” he added.
F5 Networks will deliver its “largest appliance refresh in four years,” according to CEO John McAdam. The breadth of that refresh is unclear, but McAdam revealed during F5’s latest quarterly earnings call with financial analysts (courtesy of a SeekingAlpha.com transcription) that a new 8-slot Viprion chassis will roll out within the next two quarters. Viprion is F5’s line of modular application delivery controllers. He said the new model will have double the performance of F5’s highest-performing, four-slot Viprion 4480.
An eight-slot chassis will appeal to service providers and web-scale companies (Facebook, Google, etc.). Last year F5 introduced a down-market Viprion, the two-slot 2400, which is priced for companies that want something a little more flexible and powerful than the company’s fixed-configuration BIG-IP appliances. During the earnings call, McAdam said the Viprion 2400 is selling very well and he implied that the 4480 is seeing softer sales as service provider spending has dropped across the industry.
F5 will also introduce an “application delivery firewall” via an update to its ADC firmware, TMOS. McAdam said this firewall will integrate Layer 3 through Layer 7 security, “including the loss prevention and unique application fluency to prevent sophisticated application attacks.” This firewall will also include DPI functionality, although McAdam indicated DPI would be aimed mostly at service provider customers.
NPR’s Planet Money podcast recently did an episode on Ford’s luxury car brand Lincoln, and how far it’s fallen. “Can Lincoln be cool again?” Apparently there was a time when the Lincoln Town Car was cool, finding its way into Frank Sinatra lyrics, etc. Today it’s just the ubiquitous car driven by mid-range car services. If you’ve ever walked the streets of midtown Manhattan at night looking like a tourist, some creepy guy in a seven-year-old Town Car has pulled up and offered you a ride to your hotel for a flat fee.
Ford is systematically trying to resurrect the Lincoln brand as a viable competitor to BMW, Audi, and Lexus. Unfortunately for Ford, the company has to do a lot more than simply build a good car that’s worthy of the price tag. The brand has been moribund for so long that it’s aged out of key demographics. People entering their thirties and just starting to earn enough income to think about buying a luxury car don’t think of Lincoln as an option. They’re looking at the brands that bankers, lawyers and Hollywood types are driving. Lincoln is not on that list.
I think tech companies can face the same challenge if they don’t keep their brands competitive. RIM is starting to develop a Lincoln problem with BlackBerry. Whole generations of smarpthone buyers are emerging who have never even held a BlackBerry. My two-year-old niece knows how to work the touchscreen of an iPhone or Android device. If you put a BlackBerry with a qwerty keyboard in her hands, she’d put it to her ear and say “Hello?” Then she’d throw it away and start playing with the iPhone again. That’s a heavy-handed metaphor for what’s happening with kids entering the workforce. In a BYOD world, how many 22-year-olds are going to bring a BlackBerry to work?
Postponing the BlackBerry 10 OS only compounds the problem. Some coupon web site with a silly name (CouponCodes4u.com) sent me some flash poll data aimed at gauging consumer reaction to the BlackBerry 10 news. The site polled 1,451 Americans aged 21-35. Twenty-one? How many 21-year-olds have even considered a BlackBerry?
No surprise that 29% of these people felt that BlackBerry products were not “as well designed or built” as they used to be. (Don’t young kids think that about most American cars these days?) And 59% of those surveyed said they didn’t own a BlackBerry. Why not? Well 52% of those who don’t own one cited the lack of personal and business apps, such as Instagram and Angry Birds while 53% also said that there “was nothing special” that the BlackBerry could offer them. Does that sound like a Lincoln Town Car problem? It does to me.
RIM isn’t dealing with an “If you build it, they will come” situation here. Even if they get a great OS out to the market next year and it draws rave reviews from gadget blogs, there is no market for it. BlackBerry loyalists (those few who remain) will buy one, but you’re not going to win new customers from Apple and Android. It’s going to require more than a good product. Aaaaaand it’s going to require a brand revival. And I’m not talking about easy gestures like sponsoring the pre-game show for the NBA Finals or hiring Jennifer Lopez to drive around in a Fiat. It’s going to require a fundamental invigoration of the brand. Convince those pesky millennials that BlackBerry isn’t their granddad’s smartphone.
When Cisco introduced its Open Network Environment at Cisco Live last week, Cisco executives spoke about the importance of northbound APIs, leveraging the intelligence of switches and routers to supply information to the orchestration layer so that it can make better decisions about how to program the network.
Now Plexxi Inc., a stealthy, software-defined networking start-up, is emerging with a similar message. Plexxi is being vague on the details of its technology but it’s lifted the curtain on some of its marketing message this week. It’s calling its approach to software-defined networking Affinity-Driven Networking.
Plexxi is also hinting that its software-defined networking technology extends to the physical layer of an optical network, allowing a network manager to provision bandwidth by manipulating the wavelengths of optics.
Mat Mathews, co-founder and vice president of product management at Plexxi, said his company is starting a limited private beta with select customers who have been working closely with Plexxi on the development of its products. Mathews describes Affinity-Driven Networking as an integrated software and hardware solution consisting of a a top-of-rack network switch and a software-based controller. Although Mathews describes the product as a software-defined networking (SDN) solution, he said Plexxi is not using OpenFlow.
Plexxi customers will be able to build out an entire data center network using only the company’s top-of-rack switches and its software, although customers will have the option of integrating a network domain built with Plexxi technology into an existing legacy ntwork.
Mathews said Plexxi is looking beyond the decoupling of the control and forwarding planes that other SDN vendors champion in order to focus on a neglected issue in the data center.
“Networks are fundamentally disconnected from applications and application workloads,” he said. “We can use tools and software to understand application workloads in a data center. The first part of Affinity Networking is getting an understanding of those workloads and what they require. The second part is building a physical network that can implement these workloads exactly with the direct requirements we understand from either the orchestration tools, the applications themselves or the data center operator.”
So just how does Plexxi deliver a network that is tailored to the needs of application workloads? The company isn’t sharing too much information on that, at least until its products become generally available at the end of this year or early next year. But Mathews did drop some hints.
“We’re leveraging some optical technologies to allow flexibility in how we interconnect racks together without the overhead traditionally associated with the aggregation and core layers [of a data center network]. We can connect racks directly together and enable direct east-west capacity. Servers want to talk to each other, so our physical topologies are optimized for east-west traffic. And we use software-defined networking to flexibly orchestrate that bandwidth based on where it’s actually needed.”
Plexxi’s secret appears to focus on the physical layer of the network stack. The company is virtualizing the physical layer with optical technology somehow.
“We set out to build a network where you can ‘move the wires,'” he said. “That’s where we brought in optical technology, because we can use things like wavelengths and lambdas that are provisioned in software and say, ‘OK, this rack needs to be connected to this rack. And it doesn’t just need a 20 gigabit connection. It needs a 100 gigabit connection.’ Those kinds of dynamic capacity issues are only possible where you get down to virtualizing the physical layer. Optical technology allows us to view those physical layers as wavelengths of light.”
“We can move capacity by changing the way wavelengths of light are distributed across the network,” he said. “Therefore, we can say, ‘You need 20 gigabits now. We’ll take the 10 gigabits you’re not using over here and put it there. That’s not possible today with just a [SDN] flow table.”