Finisar and u²t Partner to Acquire Exclusive Access to 100G Coherent Modulator Technology
u2t Photonics is broadening its product portfolio with the acquisition, announced earlier today, of exclusive access, in partnership with Finisar Corp., to the 100G coherent modulator technology using indium phosphide modulators developed at the former German division of COGO Optronics. The technology is based on development work carried out at renowned German research institute, the Fraunhofer Heinrich-Hertz-Institute (HHI). Modulators in InP are set to have a significant role to play in optical networks as 100G coherent and higher speed systems are deployed worldwide. Read more here.
Enterprise Cloud Adoption Held Back by Security, Cost Issues
Cloud is the biggest buzzword in technology today. Vendors large and small talk about the huge benefits to business that will come from moving applications to the “cloud,” by which they mean the “public cloud”, i.e. shared resources in a service provider-owned data center. However, despite the claimed benefits, so far few large enterprises have gone on record citing their adoption of public cloud technology.
We spoke to several large enterprises to get an update on their view of the cloud. We found strong resistance to use of the cloud on a large scale or for critical applications. A number of important issues are restraining these enterprises from buying into the cloud vision.
Our conclusion—which will be updated as we continue to speak to customers and end-users of IT—is that cloud adoption, while it may eventually go mass-market and reach even the largest enterprises, could perhaps be closer to a 10-year process than the short-term process some have suggested. That view has significant implications for many tech industry players, including the giants like HP who have claimed that cloud is the leading edge of their current growth strategy, or the VC firms investing in multiple cloud startups.
Austin Radiological Association is a major health care provider in Texas. It owns and operates 15 imaging and radiology centers, and provides services to another 17 hospitals, all in central Texas. CIO Todd Thomas is very interested in using the cloud to gain efficiencies in his IT operation. But he quickly discovered that he could not outsource many applications to service providers offering cloud services because they would not sign guarantees of data security. Health care data is tightly regulated by the federal government. To meet those standards, ARA has a Business Associates Agreement (BAA) by which a vendor commits to stand behind patient confidentiality and other government-required mandates on privacy and confidentiality. “The organizations we’ve talked to won’t sign contracts to make them liable if a breach occurs, so we can’t do business with them,” says Thomas.
Thomas’s first reaction was to rule out “Tier 1” applications involving protected patient data, and focus instead on less critical applications. Here too the going was not as smooth as he expected. He looked at moving the standard Office apps to the cloud, with either Google Docs or Microsoft’s Office 365. But once again, neither Google nor Microsoft would sign the BAA agreement, providing a legal backup for the security of the data. Thomas has moved a governance risk and compliance application to the cloud, with Compliance 360. He’s also moved an HR Information app to the cloud, with software from Ultipro. He looked at moving medical images to the cloud, but that didn’t appear viable as radiologists expect their images to appear on their computer screens in 3 seconds or less, and the potential for latency in the connections to a cloud data center rendered that option undesirable.
Moreover, Thomas has not found there to be significant cost savings from moving apps to the cloud. While there were some savings from reduced support work required of his IT staff for those apps, they were outweighed by the cost of paying the cloud providers. “In some cases, using cloud organizations to host these apps actually ended up costing us more,” he says. In theory the economies exist, but “we’d have to move tens, or even hundreds, of apps to the cloud to see any kind of significant cost savings.”
Tamas Nemeth of a nationwide logistics firm agrees with Thomas about the lack of cost savings in cloud solutions. Nemeth runs about 150 servers at his firm's data center, supporting their 1100 employees. He recently priced out cloud services with a number of providers using a cloud service provider, and also looking at pricing on website www.cloudorado.com, which aggregates cloud service provider price quotes onto a single web page based on user requirements. Nemeth’s conclusion: “The cloud might be good for unpredictable or seasonal workloads, but for a company like ours with a pretty stable, static workload, the cloud is not cheap.” Nemeth says he is focused on extending server virtualization throughout his data center with VMWare, effectively turning it into what the industry calls a “private cloud”. He says that since they run their hardware for on average five years, they get very good ROI from those investments, and he is very happy with the quality and support of the VMWare virtualization solutions. “We are getting close to having a private cloud and that is a very good solution for us,” he says. “We don’t use much cloud technology and it’s not our intention to go that way."
Henry Mayorga is Manager of Network Technology at Baron Funds, a New York-based mutual fund manager with about $19 billion under management. It’s founded and run by Ron Baron, one of New York’s best-known mutual fund managers. Mayorga reels off a string of reasons why he’s suspicious of and resistant to cloud technology. Security and confidentiality of data top the list. Like health care, money management is tightly regulated by a large body of federal rules and laws. “If someone wants to get to my data,” Mayorga says, “they need to give me a subpoena. And we have lawyers who help me decide if and how I will comply with that subpoena. If my data is sitting in a service provider’s data center, and someone shows up and demands to see my information, that service provider is likely to say ‘there’s the data, have at it’.”
Price is another important factor for Mayorga. While cloud providers claim their prices are low, he says they dictate what they will measure and how much they will charge for it. This leads to prices focused on metrics like the amount of data ingested into the cloud network (rather than the amount actually stored there). That leads to higher prices, and also difficult comparisons across vendors as each uses a slightly different formula for measuring use. Another issue is performance, and the reliability or consistency of performance. “We have had instances where our servers were sharing resources with other clients and a client uploaded 100 videos, and that affected the performance of our servers. Guarantee, or no guarantee, when something like that happens, you don’t get the performance you need.”
Finally, he cites the complexity of outsourcing to the cloud. Differing strengths and features of different providers lead to situations where different applications could be outsourced to different providers, and that leads to complexity. “I’ve seen organizations with 6 or 8 providers, and you end up managing multiple vendors with data residing in multiple places,” he says. “And then we had one service provider who had an outage that lasted four days. Had that been a critical service, I would be looking for a job.”
He adds that for a small technology startup using the cloud to develop software makes a lot of sense. But when a company has a large number of applications, and thousands of customers, the demands and requirements are just too great to risk dependencies on outsourced providers where the quality and reliability is still too unpredictable. “From a large enterprise or a Wall Street perspective, when you’ve got multiple applications and customers, it’s not a trivial problem. These are complex systems. If something goes wrong in my data center, I’ve got a shot at fixing it. If something goes wrong in a data center where a service provider has 100 or 200 customers, what chance do I have?”
Three sophisticated, experienced IT users in three large and important industries, finance, health care, and logistics each expressed serious reservations about cloud technology. The public cloud still has a lot of work to do before it can be enthusiastically embraced in the enterprise.
Companies cited in this article:
Compliance 360. Governance & risk management software (SaaS). Based in Atlanta, GA. Revenue (2010): $10.4M. Acquired by Australian firm SAI Global (ASX: SAI) in 2012.
Ultimate Software. Finance & HR software. Based in Weston, FL. Public (Nasdaq: ULTI). Revenue (2012) $332.3M. Net income $14.6M.
VMWare. Virtualization software. Based in Palo Alto, CA. (Nasdaq: VMW) Revenue 2012: $4.61B. Net income: $746M.
Global Upside announces support for its client Zillow. Based in Los Altos, CA, Global Upside is a leading provider of outsourced finance, accounting, and HR services worldwide. Seattle-based Zillow is the leader in online real estate information, serving both real estate professionals and consumers. Zillow recently reported a record quarter with outstanding growth in revenue and web traffic. Global Upside provides scalable support for fast-growing companies like Zillow.
More details in the press release here.
About Global Upside:
Headquartered in Los Altos, California, Global Upside is a specialist in finance, accounting, and HR services. Global Upside’s team of finance professionals delivers high quality, timely, and responsive service to clients in more than 40 countries worldwide. For more information, visit www.GlobalUpside.com.
Gazettabyte Looks at Innovative Optical Components from u²t Photonics
Roy Rubenstein of Gazettabyte looks at new products from u2t Photonics for high speed optical networks. A new balanced photodetector can receive light signals at up to 1 terabit per second (Tb/s), to enable the next generation of high speed optical networks for ever-expanding Internet traffic flows. u2t is also introducing new products for the transmit side of the optical network, including new high speed modulators. These new products will be used by optical system makers and test and measurement companies worldwide.
u2t Photonics pushes balanced detectors to 70GHz
u2t's 70GHz balanced detector supports 64Gbaud for test and measurement and R&D
The company's gallium arsenide modulator and next-generation receiver will enable 100 Gigabit long-haul in a CFP2
"The performance [of gallium arsenide] is very similar to the lithium niobate modulator"
Jens Fiedler, u2t Photonics
Read Roy's article here.
About u²t Photonics:
u²t Photonics AG, a privately-held company headquartered in Berlin, Germany, is a leading supplier of innovative optical components up to 100 GHz. On the basis of its unique and mature technology, u²t develops, manufactures and markets highly competitive products for high-speed communication applications as well as detectors with superior performance for T&M equipment vendors. Being ISO 9001 certified and providing high quality service, u²t offers a comprehensive and fully TELCORDIA qualified portfolio.
Exponential Interactive Calls on VPTax's "Tax Director Down the Hall" to Help Plan Next Growth Phase
Exponential Interactive, one of the leaders in the digital advertising revolution, uses VPTax's outsourced tax director model to get more knowledgeable, efficient tax guidance than it could get either from in-house staff or a large accounting firm. Further details in the case study below.
Exponential Interactive is a pioneer in digital marketing. Based in Emeryville, California, and active in 26 countries, Exponential has pioneered the application of intelligence to Internet marketing and advertising to help major brands target and reach consumers. Exponential is heading towards an IPO
and in 2012 filed an S-1 document showing that in 2011 it earned net income of $6.8 million on revenue of $169 million. Unusual for a technology startup, Exponential has been profitable for the past 11 years.
In 2011, Exponential Chief Financial Officer John Rettig retained VPTax to manage Exponential's tax affairs. Rettig, who has been Exponential's CFO since 2005, was considering bringing on a full-time Tax Director, but after meeting the VPTax team, he was persuaded of the value of using VPTax to provide part-time Tax Director services. The part-time model enables Exponential to take advantage of the broad and deep corporate tax experience of the VPTax team, but at a lower cost level than the salary of a full-time staff Tax Director.
"For Exponential, an outsourced Tax Director is the right model," Rettig explains. "VPTax helps us get our heads around complex tax issues, and their support has enabled us to go from being reactive to being proactive on the tax front."
According to John Rettig, many smaller growth companies could benefit from the VPTax outsourced Tax Director model. "Too many companies don't think about taxes until it's too late to take full advantage of the opportunities."
VPTax brings extensive experience of corporate tax issues. Todd Suchevits, the VPTax principal who serves as Exponential's outsourced Tax Director, has more than 20 years of corporate tax experience, including extensive experience of international tax issues and structures. With Suchevits' guidance, Exponential was able to proactively manage the diverse legal and audit firms providing tax advice and find solutions that positioned Exponential well, while at the same time keeping costs and complexity under control.
Many accounting firms have a "public accounting" culture, often not conducive to rapid decision-making. Most of the VPTax team have worked on the corporate side as well as in public accounting, giving them a useful perspective on the needs and challenges of managing finance for a fast-growing tech startup like Exponential. "VPTax have been on the other side of the fence, so they know how to do things efficiently and not waste time," says Rettig. "With VPTax, we're able to make what outside advisers do more efficient and less costly."
According to John Rettig, many smaller growth companies could benefit from the VPTax outsourced Tax Director model. "Too many companies don't think about taxes until it's too late to take full advantage of the opportunities," he says. "When a company is losing money early in its life, that's the time to start thinking and planning for the longer-term tax situation. VPTax has unique experience and expertise and can be a valuable asset in that process."
u²t Photonics Unveils the World’s Fastest Balanced Photodetector with 70 GHz 3dB Bandwidth
A significant step forward in the race to upgrade the capacity of fiber-optic networks to cope with the deluge of mobile and landline internet traffic…
u²t Photonics Unveils the World’s Fastest Balanced Photodetector with 70 GHz 3dB Bandwidth
Berlin, Germany, May 21, 2013 – u²t Photonics AG, the provider of leading-edge optoelectronic components, today announced the world’s fastest balanced photodetector offering a 3dB bandwidth of 70 GHz. The
device supports systems for next generation networks using 400 Gbit/s or 1 Tbit/s coherent detection-based optical transmission.
The BPDV3120R complements u²t's product portfolio of ultra high speed photodetectors and receivers. The optical frontend consisting of a monolithic balanced photodetector chip with on-chip biasing has been successfully tested in a system environment and will be used in Test & Measurement (T&M) equipment, such as optical modulation analyzers, for the test of next generation coherent network components and modules.
The coaxial single-ended output can detect up to 64 GBaud polarization diversity x-QAM signals featuring highly reliable results for common mode rejection ratio, linearity and optical input power. Therefore, the device can be used in next generation long haul transmission systems at data rates of 400 Gbit/s and beyond. At the same time, the BPDV3120R is perfectly suited for T&M applications as well as ongoing R&D activities at high Baud rates.
“While 100G coherent systems are being widely deployed today, we can already see that 400 Gbit/s or even 1 Tbit/s systems will be required in a few years to support the continued exponential growth of data traffic in mobile and fixed networks,” said u2t Photonics CEO Andreas Umbach. “Our 70 GHz balanced photodetector provides one of the key building blocks that will enable systems providers to build and deploy systems operating at 400G per wavelength or higher.”
u²t started sampling the BPDV3120R in March 2013 and will start its production in June.
u²t Photonics AG, a privately-held company headquartered in Berlin, Germany, is a leading supplier of innovative optical components up to 100 GHz. On the basis of its unique and mature technology, u²t develops, manufactures and markets highly competitive products for high-speed communication applications as well as detectors with superior performance for T&M equipment vendors. Being ISO 9001 certified and providing high quality service, u²t offers a comprehensive and fully TELCORDIA qualified portfolio.
An excellent new report from InformationWeek entitled "2013 State of Cloud Computing" sheds some light on one of the tech industry's best-kept secrets: use of "the cloud" by business (specifically, public cloud services) has yet to be embraced by corporate America and continues to grow slowly, more "tortoise than hare" as the report says. While public cloud growth has been slow so far, it will probably take off at some point in the future. Meanwhile, the report and its survey results contain valuable lessons for players in the cloud business, including tech vendors, end-users, and the press and analyst community that writes about the cloud.
Based on a survey of 446 IT professionals at companies in the multimillion and billion-dollar range, the InfoWeek report concludes that the cloud is seeing "tepid adoption that's both unsurprising and discouraging." Unsurprising to the author perhaps, but surprising to many of us who read the incessant trade press and analyst hype. Asked about their plans to move corporate services to the cloud over the next 24 months, the largest group of respondents, 35%, said they foresee just 10% to 24% of their IT services delivered from the cloud. Combining them with the group who said they anticipated "very limited usage" of cloud (26%) and those who said "none, we hate the cloud" (5%) yields a total of 66% of respondents seeing none to modest use of the cloud over the next 24 months. What's more, this share has barely shifted from the previous year's survey. Private cloud (i.e. server virtualization) in contrast sees far higher adoption rates.
The most interesting lessons from the InfoWeek report for tech vendors come from the cloud concerns and handicaps identified by respondents. The first handicap, named by 51% of respondents, was "security defects in the technology itself." After that came the risk of "unauthorized access" to confidential information, an issue for 45% of respondents, followed by concerns over the performance of cloud-based applications (31%), concerns over business continuity (30%), and finally concerns over the viability of the cloud provider's business (30%). The percentage voicing those last two concerns actually rose between 2012 and 2013. This fairly intimidating list of perceived weaknesses provides a clear road-map of issues and concerns that tech vendors need to address, and address with objective evidence and data, to demonstrate that their individual products and services do not suffer from the concerns over security, confidentiality, performance, and reliability. Some vendors may actually be able to demonstrate that their technology solves some of these problems. One reason I am bullish on WAN optimization technology is that WAN (i.e. Internet) bandwidth is an important impediment to getting good performance from a cloud application, so WAN optimization, whether via an appliance or through the cloud itself, can not only deliver ROI itself, but also open the door to much wider use of the cloud.
An example of a strategy that flies in the face of these results might be that of Amazon Web Services, which apparently believes that aggressive price cutting will drive a "revolution" in IT and bring them the thousands of enterprise customers they hanker for. If the InfoWeek report is accurate, price may be largely irrelevant to large-scale enterprise penetration of the cloud.
End-customers, such as service providers contemplating an entry into the public cloud business, can learn from this report by figuring out how they can position themselves as having overcome all the problems identified in the InfoWeek survey, which the existing cloud provider pioneers are suffering from today. Another important lesson of the study is that the majority of respondents "don't see the point" of automation and orchestration within a cloud implementation. So effective, reliable automation is clearly a feature that could raise customer receptivity to public cloud solutions.
Finally, press and analysts should ask themselves if they are really providing value to their readers by merely reiterating and adding to the tidal wave of "cloud revolution" hype generated by vendors. An independent perspective is more valuable than a chorus of cheerleaders. I have no doubt that at some point in the next five or so years the public cloud will become a disruptive force among large enterprises as VMWare CEO Pat Gelsinger said on Wednesday, and as it already has for consumers who buy music, but meanwhile there may be better places to invest time, money, or technology.
The Decline of Business Journalism—Jim Cramer Tells It Like It Is
CNBC personality and Street.com founder Jim Cramer highlighted a sad but important trend in a speech earlier this month: the decline in the quality of US business journalism over the past decade or so.
The decline in the volume, breadth, and quality of business journalism is a result of the rise of the Internet, which has decimated the revenue of so many publications. Today, as Cramer pointed out, thousands of public companies do not get any independent, critical press coverage, a radical change from 20 years ago when local newspapers, local TV stations, and national services like Bloomberg had the staff resources to cover small and midsized companies regularly. “The press used to play a vibrant role in trying to cover companies in an objective and at times investigative way. That’s out the window [today],” said Cramer.
In my opinion, the situation is even worse in tech. The quality of the tech trade press has declined as fewer reporters have been forced not just to write more articles, but also to post, tweet, and record videos every time they do a story. More activity, more page views, but less information-gathering and less thought or analysis. The local business press (metro newspapers) has pretty much given up (as Cramer points out, cities thousands of miles from New York have fired their business staff and simply buy in their business coverage every morning from TheStreet.com or other such organizations.) On top of those trends, in tech there is a new breed of online publication that is either owned by tech companies or edited by tech “fanboys” providing cheerleaderish coverage with little perspective.
Ironically, today, despite billions of hits for any search you care to do, it’s much harder to know what’s really going on with small or medium-sized companies than it was 20 years ago. Companies need to create their own communications—and rise to a higher standard than before. Many companies have not risen to this challenge—often their communications staff parrot the jargon-ridden statements of their top executives, which make sense only to the small group of people who already know them. It’s often technology for its own sake, with little context or perspective. What these top execs need are communications staff who can bring some insight on what the marketplace wants to know, instead of merely reaffirming their hype. A good example might be Ciena’s blog. This blog is often more interesting than the company’s press releases. And usually more informative than anything the sell-side analysts write about the company (another, separate, tale of quality decline that Cramer mentions in his speech).
As for journalism, I agree with Cramer’s recommendation, which is that more publications follow the example of the Wall Street Journal and NY Times and move to paywalls. “The propensity for free journalism is a bad one. We were all taught that professionals should never give away their product for free,” Cramer said in his speech. Paywalls may not apply in some areas of online journalism (such as celebrity news), but for those areas where your information adds real value, the way of the future is to figure out how to charge for it.
Incidentally, in the speech, Cramer showed himself to be as good a journalist as he was a hedge fund manager. And for any journalists reading this, hedge fund managers can make some of your best sources—Cramer was one of my best sources when I covered networking for Bloomberg News. Unlike VCs they are truly independent (they can see the bad as well as good side of their holdings) and unlike sell-side analysts, they actually do their homework.
Jim Cramer: “The propensity for free journalism is a bad one”
Nuage Networks launched last week with an SDN launch that takes us significantly closer to real-world deployment of SDN than ever before.
Nuage, a subsidiary of Alcatel-Lucent, described its solution as “second-generation”, trying to differentiate it from the first wave of SDN companies that have talked mostly about architectures and paradigms and revolutions and not enough about specific use cases or product shipments.
Interesting aspects of Nuage include:
· Nuage’s SDN solution, named VSP for Virtualized Service Platform, seems aimed primarily at networking within large data centers (aka clouds). Those data centers could be operated by public cloud providers or enterprises.
· Nuage claims VSP will cut through the operational bottlenecks in today’s data center networks, speeding up configuration and provisioning times from “days to minutes”, particularly as virtual machines are created or moved around.
· Nuage’s VSP offers features of L2 through L4 networks, differentiating it from some other SDN solutions that focused primarily on the L2 challenges of data centers. To support this, Nuage included quotes from F5 Networks and Palo Alto Networks, stating that they are working with Nuage to bring, respectively, L4-7 applications and firewalls to cloud networks. This raises once again questions about how SDN might impact the business model for standalone higher-level networking services and appliances.
· Refreshingly, Nuage’s press releases included dates and customer names. Specifically four customers will be trialing VSP beginning this month. They are Canadian service provider Telus, French wireless service provider SFR, British cloud service provider Exponential-e, and American health care provider University of Pittsburgh Medical Center (UPMC). UPMC is quoted saying it expects the Nuage solution to help it further its strategy of virtualizing its data centers.
· VSP will begin commercial shipments in “mid-2013”.
· The Nuage press release and White Paper are conspicuous in the absence of any mention of OpenFlow. Martin Casado, founder of Nicira, which is also focused on supporting intra-data-center networks, has gone further and said OpenFlow does not have a role to play in intra-data-center network virtualization. OpenFlow is however supported by dozens of companies who have joined the ONF, ranging from giants like Cisco to small switch manufacturers. Its future will be interesting to watch.
· With its software-based solution, Nuage clearly places the emphasis of SDN on automating network functions and services, to bring higher-level applications into software and into the data center. The implication is that the “revolution” of SDN will impact higher-level application vendors much more than it may impact Layer 2 switch vendors. That is not what the Wall Street community has focused on, where the excitement, such as it has been, has been more about the potential “commoditization” of switching and what that might mean for the big switch vendors.
On Monday, Margaret Thatcher, one of the most influential political figures of the 20th century died, aged 87. Palomar’s founder Jeff Ferry remembers one of his encounters with her below.
My Favorite Memory of Margaret Thatcher
I was a young TV reporter covering the new Conservative Government in 1979 when Margaret Thatcher attended her first European Union summit at Dublin Castle.
Back then it was called the European Community, and there were only 9 members, as opposed to 27 today, but what has not changed is that they could almost never agree on anything and in the rare cases they did agree on something, it usually did more harm than good.
At this particular Summit, enormous curiosity surrounded the first-ever female PM of a British government. The Irish government had invited her to stay in Queen Victoria's old bedroom in Dublin Castle as a nice token of respect--and a sign that 800 years of colonialism was mostly forgiven and forgotten. Nevertheless, Maggie threw the Summit into instant uproar by her intransigeant demand to get a full "refund" of Britain's one billion pound contribution to the European Community budget. Night after night, the Summit talks went on behind closed doors as the other heads of government tried to persuade Maggie to compromise.
We journalists gathered in various reception areas in the Castle to wait, munching on hors d'oeuvres, consuming Guinness, fine German lager, or French wine kindly provided by our Irish hosts, and idly discussing politics, upcoming elections, or national differences among journalists. It was a source of mystery, for example, why Italian journalists were the best-paid. It was no surprise that they were also the best-dressed. French journalists were of course the most arrogant. In those union-dominated days, British TV was notorious for the size of its team (I had a six-man crew with me to shoot interviews, while most of the other nations had crews of two or three). Every hour or so, some official would come out of the meeting room, we would gather around him or her and ask for an update. The report was always the same: "The British are still demanding a refund. The other eight are not accepting this. The British are isolated." Finally, at about 4 in the morning, Maggie herself came out. We mobbed her instantly. The foreign reporters let the Brits do the questioning--they knew we would be more aggressive and less respectful than they. One of the veteran print reporters led the attack:
"Is it true Prime Minister that you are holding firm on your demand for a billion? Is it true that the British are completely isolated?"
Maggie gave him that look that I would come to know and love for years after: she narrowed her eyes for an instant as she sized up this person whom she didn't much like, followed immediately by a beaming smile to show him and the audience at large that she was unruffled and serenely confident she would ultimately triumph.
"You are absolutely right!" she replied enthusiastically. "The British government is sticking to its guns and demanding its billion pounds back, as is its right. The other eight governments are completely isolated."
We British journalists immediately got the joke. Some smirked, some laughed, but we all noted it in our notebooks. We gave the foreign journalists a few seconds to comprehend her comment, and watched them too break out in smiles as they got her joke. She held still and beamed her Maggie smile the entire time, as if she were posing for a portrait.
And so it went on for the next twelve years. She did what she thought was right, whatever the opposition. And she made some great changes to a country that in 1979 was mired in defeatism and negativity. The journey isn't over but it began in 1979.
PM Maggie Thatcher in Moscow, 1989. Thanks to Baltimore Sun for nice photo essay. For another interesting take on Maggie, check out this column by Andrew Sullivan.
High-Speed Optical Networks: Breakthrough Modulators in New Material
With escalating data traffic on the world’s networks, the pressure to carry more bandwidth never ceases. For optical networks, 100 Gbit/s wavelengths are in the early stage of mass deployment, but component companies are already looking at the future, to 400Gbit/s and beyond.
At the recent OFC conference in Anaheim, U2t Photonics, a leading component maker headquartered in Berlin, Germany, presented results of tests on a breakthrough new modulator, made from gallium arsenide (GaAs), an alternative material to lithium niobate, the standard material for most modulators in use today. The GaAs modulators tested were fabricated in an array of six modulators on a single chip, capable of transmitting up to 440Gb/s via the single array.
The paper, titled Fabrication of the First High-speed GaAs IQ Electro-optic Modulator Arrays and Applicability Study for Low-Cost Tb/s Direct-Detection Optical OFDM Networks, by L. Stampoulidis et al, was written by u2t researchers based in Germany and Britain in partnership with academic researchers from research centers in Germany and Greece.
A promising technology for the future.
More info available at:
http://www.ofcnfoec.org/home/program/agenda-and-abstracts/ (password required to access full paper)
SDN, Service Providers, and the Prisoner’s Dilemma
SDN was the leading subject of conversation at OFC 2013. But beneath all the stimulating talk and exciting architectural diagrams, there are varying agendas and objectives.
Stanford Professor Nick McKeown laid out the core argument for SDN technology, and an impressive lists of its benefits and improvements over
today’s architecture. As he explained in his plenary, SDN will free networks from vendor lock-in and enable software apps to run in the network, just as you can download an app from the App Store and let it make your phone do a thousand different things.
Figure 1 below shows how we might move from a vertically integrated vendor-dominated network to a more open and flexible network architecture. Chipmakers like Broadcom are already working on switching chips optimized for OpenFlow; and apps that could ride on top of an SDN network are being talked about by large companies and startups. For the control plane layer, there were ideas, diagrams, and conversations aplenty. But so far, real products and deployments are still to come.
The service providers at OFC were vocal in their support for SDN. For them, much of the advantage comes from the relatively simple idea of having one common software layer to manage both the IP and the optical layer. (see Fig. 2 from Google’s presentation). Optical vendors like Ciena, Infinera, and BTI talked about their support for SDN. Adva engineers, in partnership with Telefonica of Spain, gave a paper about their trial of an SDN-like “middlebox” controlling both the optical and IP layers (See Fig. 3).
Watch What I Do, Not What I Say
However, how real is optical vendor support for the SDN vision of one common control plane? If that common control plane existed, it would make it easier for service providers to substitute competing optical systems for each other anywhere in their network. That would give them greater power to drive down optical system pricing. As CenturyLink SVP Dr. Pieter Poll argued in his keynote, optical prices need to come down even faster than they have in the past several years to help service providers pay for the investment to carry more traffic. And while router vendors say they support SDN, do they really want to give service providers the freedom to transfer more traffic out of the IP layer, so reducing their use of expensive and power-hungry routers?
The actors in this three-cornered drama are like the prisoners in the famous prisoner’s dilemma. What they say may be different from what they do, and what they do may ultimately not be in their own best interests. Clearly, for service providers, the best end-result is for no single optical vendor to dominate the market, so service providers have the clout to use their buying power to get what they want. That’s not the actual situation on the IP side, where one vendor does dominate and doesn’t always listen to customers whose requests it finds unpalatable. Perhaps as a result, service providers have tended to favor optical in recent years (see this comment, suggesting that service providers are spending 4X more on optical than routing today). But despite growing demand, optical vendors still suffer sub-optimal rates of return and poor gross margins. Further downward pressure will drive the industry away from those markets that demand high returns (the US), and towards those markets that accept lower return (parts of Europe and Asia).
One company doesn’t suffer from divided interests: Google. It builds virtually everything itself. That may be one reason why it stands out as the only clear example so far of a successful SDN deployment.
Figure 1: Transition from today’s proprietary networks to the open world of SDN. (Source: Prof. McKeown, OFC 2013 Plenary talk)
Fig. 2. Google’s vision of one SDN-based NOS controlling the entire WAN.
Fig. 3. Telefonica/Adva/CRAAX Lab (Spain) offer a vision of a “Middlebox” that provides analogous functionality to SDN. Trialed successfully in the Telefonica network. (Source: Coordinated Operations between IP and Transport Network Management Systems in Multi-vendor Settings, M. Yannuzzi et al. OFC 2013)
One of the key issues we hear from clients is “how can I plan my SEM budget?” The answer undoubtedly varies but the key is to be smart about it. When it comes to search engine marketing, you have to pick and choose your battles.
You have to realize that you may not be able to do everything you want to right away. It can be expensive to get a SEM campaign off the ground, and you can’t always afford to do it all at once. You have to decide what it more important. Do you want to up the frequency of your company newsletter to connect with customers, or do you want to allocate more of your budget to bidding on more competitive PPC keywords? The first step is to start with a list of goals to help you decide where you should start and how much money you should dedicate to each effort.
Second, you need to understand what your competition is doing in terms of search marketing. Where are they focused on and where are they lacking? If they are dominating the search results, you know you’ll have to devote more time and energy to SEO in order to compete. If there is no clear leader in your niche, then a smaller amount of money in the right place is going to make a noticeable difference. Big brands tend to have the money and manpower and own highly competitive keywords, both in the organic search results and PPC advertising. But what keywords can you focus on that they’ve missed?
It’s important to understand that SEO and SEM are two very different things. One of the key differences is that SEM involves PPC, a paid marketing tactic. Most SEO tactics (link building, social media marketing, article marketing, etc) don’t cost anything, but require a serious time commitment. What do you have more of and what are you willing to spend? PPC might cost you more, but can lead to faster results for as long as you’re running ads. SEO is a long-term process that builds on itself over time. So, if you want the best of both worlds, put simply, your site can benefit from both as there are numerous advantages of appearing in both paid and unpaid listings.
Around 55-65% of users will click on the organic search engine listings, and around 35-45% will click on the paid search engine results (depending on whose research you believe). Engaging in both SEO and SEM allows search engine users to be ‘captured’ at all stages of the buying process. Users that are in the problem recognition or information search stages of the buying process are more likely view organic listings. On the other hand, users who are closer to making a purchase decision or are evaluating alternative products and services are more likely to click on SEM advertisements.
Silicon Photonics Generates Excitement at OFC 2013
The topic attracting most of the talk at this year’s optical trade show, OFC/NFOEC, was silicon photonics, the technology of using silicon for optical communications inside computers, between computers inside data centers, and in long-distance telecom networks. Hot new startup names include Aurrion, Kaiam, Kotura, and Skorpios.
Silicon photonics is important because traditionally optical networking is done with expensive compound semiconductor materials like indium phosphide (InP). Silicon is not only cheaper than compound materials, but more importantly, manufacturing techniques for silicon chips are much more mature, allowing much higher yields per wafer and so lower cost per chip.
Most of the talk at the show was not about doing new things in silicon, but finding new ways to fix traditional compound semiconductor lasers onto silicon chips—important breakthroughs in manufacturing techniques.
Kotura announced a QSFP 100Gig module aimed at data center networks at OFC. It also announced partnerships with high-speed electronic chipmaker Mindspeed and BinOptics, a private maker of lasers. The lasers are made of InP, but bonded onto the silicon chip, using an etched facet technology that amplifies the power of the light on the silicon chip. Kotura claims that the small size of the QSFP will enable switch makers to fit 40x100G of bandwidth or 4 Terabits onto the front of a switch. BinOptics claims to have shipped 40 million lasers, while Kotura recently said it is shipping 60,000 channels per month and has over 130 patents filed or granted. Kotura might be looking at an IPO sometime soon.
Other private companies leading the way in silicon photonics include Aurrion, Skorpios, and Kaiam. Aurrion said in a post-deadline paper that it is able to make both 100G data center chips and 100G telecom chips from the same wafer using a material transfer technique that produces a high-yield, low-loss optical contact between the compound semiconductor material and the silicon substrate. Aurrion said it's able to implement 16 lasers, each operating at 25 Gigabits/sec, for total bandwidth of 400 Gigabits/sec, all on a single silicon chip.
Fig. 2. Telecom and datacom (16 lasers) from the same silicon wafer. Source: Aurrion.
Meanwhile, Cisco is expected to bring out silicon photonics-based client modules on its 15454 metro optical system later this year. In recent months, those rumors have hit Finisar’s share price as investors worried that Finisar, which has no silicon photonics solutions today, could lose market share with its most important customer. Finisar stock declined to the 13 ½ range during OFC, despite investor and analyst chitchat that optical demand should pick up in the second half of this year.
Seems like some of these hot Si photonics startups are likely to get acquired by larger component vendors over the next 12 months.
This weekend GigaOm Pro published a new report on software-defined networking (SDN), forecasting that the market for SDN products and services could grow from "under $100M" last year to $340M in 2014 and $2.45B in 2018.
Meanwhile, over in the real world, the signs are growing that SDN is losing its luster and that last year's "hottest networking technology in a generation" is in danger of becoming this year's cold soup that was left too long on the stove.
The woeful lack of customer deployments of SDN networks is very disheartening. A year after Nicira launched, and more than two years after the creation of the Open Networking Foundation, there is scarcely a public customer announcement to be found. The five impressive customers (led by Fidelity) who said they were deploying Nicira network virtualization a year ago have not to my knowledge spoken publicly about how they are using that technology and what benefits it might be delivering if it was in fact ever deployed, outside of a test bed.
Last week I met with a senior engineer at one of the world's largest service providers. This service provider has been a vocal supporter of SDN and I asked him for his view as a leading proponent. "Wait a minute", he said. "Just as there is a lively public debate, there's a lively internal debate at our company. Many of us think it will take many, many years for SDN to reach maturity." He pointed out that the exact same idea as SDN, unified software control of the control plane, was tried in the optical layer a decade ago with GMPLS, and that went nowhere. He was clearly more interested in the latest innovations from Cisco and Juniper than in anything happening among the crowded field of SDN startups.
The startups trying to position themselves as SDN plays are looking increasingly tarnished, producing an endless stream of buzzwords and simplistic analogies, illustrated by their use of simple-minded irrelevant videos that remind one of the priceless moment in Trading Places when the old white guy commodities traders show Eddie Murphy a glass of orange juice to help him understand the market for orange juice futures.
Big Switch Networks' recent launch was a good example of the becalmed nature of the SDN "market". In Q4 2012, Big Switch changed out its marketing team and then launched three new products, amidst what passes for great fanfare in the networking market. Lacking customers, Big Switch re-announced their broad swathe of partners. Indeed Big Switch has announced more partners than Taylor Swift has had boyfriends, but like the unlucky-in-love Ms. Swift, Big Switch seem unable to consummate a single relationship. Partners who don't bring customers are like a boyfriend who won't buy you dinner. The buzzwords about "exploding markets," "affinity networks," and "networking revolution" begin to ring a bit hollow. One begins to suspect that all these startups, their strings being yanked by VC owners, are simply trying to stampede the nervous tech giants into following VMWare and paying big money for a technology promise that may (or may not) represent the future. Today, Big Switch was able to announce that Intel Capital has put $6.5M into their pot, bringing their total raised to $45M. But as the WSJ pointed out, Intel has an interest in backing anything with a chance to put more servers into data centers--and its track record in networking is uninspiring.
I recently spoke to a friend who runs a university network. He said to me: "I'd love to test SDN in my lab. I'd need about six switches and a controller. But I can't find anyone to help me." The startups need to get smarter with their product development, marketing, and customer outreach.
Unlucky in love? Sometimes the solution is not more lipstick.
An excellent article from an IT-economics think tank in Washington shows why manufacturing, including technology manufacturing, is so important to US prosperity. Entitled Worse than the Great Depression: What Experts Are Missing About American Manufacturing Decline, it is written by four economists at the Information Technology and Innovation Foundation (ITIF). To quickly summarize its key findings:
· In the ten years 2000 to 2010, the US lost 33% of its manufacturing jobs, greater than the 31% manufacturing job loss experienced in the Depression years of 1929-33.
· In the 2007-2009 recession period, the US lost 16.3% of its manufacturing jobs, greater than the percentage loss in any other recession since WWII.
· Normally a steep downturn is followed by a sharp pickup. But not this time. In the 30 months following the end of the recent recession, the US only increased its manufacturing job base by 0.7%, leading the authors to describe manufacturing as stuck in a “one-way job loss rachet.”
· Beginning in 2002, the US began to run a trade deficit in advanced technology products, including life sciences products, medical devices, optoelectronics, information technology, and aerospace—“the very products in which the United States is supposed to have a competitive advantage,” the authors point out.
· This manufacturing decline is an important reason why between 2000 and 2010 the US lost 40% of its jobs in computers and electronic products.
There are a lot of economists and politicians arguing that the US is transitioning to a services economy and that services are just as good as manufacturing. Here in Silicon Valley, there are lots of people claiming that the US can prosper by inventing “apps” and other pieces of software, and leave the nasty work of actually making things to others. “We are an ideas economy,” claimed Larry Summers. See the graph below to see what an ideas economy looks like.
For more than 200 years, manufacturing has been the world’s most important engine of economic growth—because it’s the only sector of the economy where productivity can steadily increase by a significant percentage each year and at the same time deliver widespread income growth to a large workforce. This is due to two key features of manufacturing: first, it has large upside leverage, because manufactured goods can be exported worldwide, and their markets grow as productivity improvements drive down the price; and secondly because manufacturing requires a workforce that grows roughly in proportion to the output of the product. While the services industries have advocates claiming the same benefits for services, if you think about it you can see it’s usually not the case. Take Facebook. This is a service business that has seen dramatic growth in its user base to some 1 billion users worldwide and some $4 billion in annual revenue. But all that has been achieved with a relatively small (approx. 3,500) employee base. This is roughly an order of magnitude smaller than Sun Microsystems, the company that previously occupied the sprawling Menlo Park campus which Facebook is now busy trying to fill and also upgrade by spending its billions on grand architectural plans. The tech lobbying group TechNet and others have tried to argue that FB is part of an “app economy” which is generating billions for California. This is to an extent true, but the fact remains that those billions go primarily to the shareholders in those organizations, and secondarily to the relatively small number of employees there. There are very limited multiplier benefits to the larger regional economy.
Then there is the other type of service business, the low value-added service business such as a retailer or a fast food restaurant. They do employ large numbers of people, but (a) they pay very low wages and (b) productivity growth does not typically lead to rising employment. On the contrary it leads to falling employment, because it usually entails replacing staff with technology. These low-tech services businesses cannot export, so the only way to increase productivity is to reduce staff.
The graph below shows the unemployment rate over the last 20 years in Santa Clara County, the heart of Silicon Valley. In that time, the Valley has nurtured the renaissance of Apple (world’s most valuable company), Google, and Facebook. And yet unemployment has continued its relentless long-term increase. The long-term secular trend in the graph below shows it rises on average of 0.2% per year and has doubled in 20 years, from 4% to 8%. And this is one of the most prosperous regions in the US!
As the ITIF authors state: “It is impossible to have a vibrant national economy without a globally competitive traded sector.” By “traded sector” they mean a sector capable of exporting large volumes of goods.
So what can be done? I will look at that in a future post.
Monthly unemployment rate, Santa Clara County.
Data from BLS via the St Louis Federal Reserve Bank.
Trendline (red) added.
Money Makes The World Go Round @palomarcomm - Tumblr Blog | Tumgag
