Koruta, which I profiled last November, makes a photonics chip that allows signals to pass between chips using light (photons) instead of electrons. This makes communications between chips faster, something becoming more important inside the data center as networks become flatter, faster and fatter.

The Kotura chip is a fiber-based transceiver that can deliver 100 gigabits per second inside the data center. The transceiver could live on a board next to the CPU or inside a switch and could eventually expand to deliver a terabit per second (Tbps). Currently it’s used in high-performance compute clusters, which are also the most-popular home for Infiniband. From my post in November:

While one 1 Tbps is crazy fast when you consider that many data centers are currently upgrading to 10 gigabit Ethernet between servers, it’s going to be necessary. Arlon Martin, VP of Marketing, Government Contracts & Industry Relations at Kotura, tells me that customers are building products for the high-performance computing sectors but also for real-time data processing. The goal is bringing a low-power and less expensive optical part into a rack of servers, able to scale up to terabit per second capacities.

Kotura isn’t the only company trying to bring fiber optics into the data center. Plexxi is building fiber-based switches, while Facebook and the Open Compute Project in January announced their own plans to integrate photonics into their open hardware program. Intel, Cisco and IBM all have research or have acquired startups in this space as well. With its emphasis on fast networking, it makes sense for Mellanox to follow suit.

The company plans to continue offering the Kotura transceiver and open a research center in Monterrey Park, Calif., where Kotura is headquartered. From the Mellanox release:

“Operating networks at 100 Gigabit per second rates and higher requires careful integration between all parts of the network. We believe that silicon photonics is an important component in the development of 100 Gigabit InfiniBand and Ethernet solutions, and that owning and controlling the technology will allow us to develop the best, most reliable solution for our customers,” said Eyal Waldman, president, CEO and chairman of Mellanox Technologies.

At the time of my profile, Kotura had raised undisclosed millions from ARCH Venture Partners, Fuse Capital, GF Private Equity and others. It has an established customer base in the telecommunications business where it has sold product since 2006. But last year it began targeting the data center, where it apparently attracted Mellanox’s eye.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Infrastructure Q4: Big data gets bigger and SaaS startups shine
• Infrastructure Q2: Big data and PaaS gain more momentum
• Big Data, ARM and Legal Troubles Transformed Infrastructure in Q4

But that doesn’t mean you can stop worrying about hardware redundancy or high availability data centers. As cloud applications get more complex, you must add complexity at the hardware and infrastructure level. The key is that complexity will be under your control, which will help your application stay resilient.

Resiliency isn’t just a buzzword anymore

The very complexity of modern distributed or even vertically integrated solutions (I.e., Amazon vs. VCE vs. legacy IT architectures) means that you can’t accommodate all risk vectors. Adrian Cockcroft of Netflix is a brilliant cloud use/application design strategist, and he has many other incredibly skilled people working with him, yet Netflix was still unable to account for every variable that might occur in the complex system (Amazon) that supports his complex system, causing an outage affecting millions of customers on Christmas eve.

I’ve been designing, building, implementing, and operating critical infrastructure for enterprises for over twenty years now. In that twenty years I’ve uncovered a large number of truths about the IT world, and one of those truths is “you can’t assume anything”. Every time a server is installed, a switch configured or a new update applied to an application a change has occurred. The change might have been tested under every known scenario (impossible), yet still fail because of a heretofore unimagined use case. In my experience, it’s often the combination of activities that occur at the worst possible moment that cause you problems.

Go hybrid, young developer. Go, hybrid.

“ <a title="Attribution-NoDerivs License" Photo courtesy of Flickr user piermario

There has been considerable discussion about resiliency lately, and in some cases the underlying concerns around management and performance of complex systems has been well documented. One of the best discussions on the topic of resiliency was by Richard Cook.

Mr. Cook points out, as I’ve written in the past, that your operations practices and people are often the keys to your success. What I would also argue is that when you consider your design, locations, people and process you should include strategy for system reliability beyond mere application resilience.

Many of the companies I work with have global organizations with tens of thousands of internal and or external customers. In all cases these companies have decided that one cloud isn’t the right approach. A hybrid approach, while more complex in the beginning, offers significant upside in several areas over the long term besides higher reliability, including avoiding lock in, and having the ability to deliver service to any region, regardless of the available cloud provider. These same companies have also decided that carrying the risk that their most important IT assets are down through acts of god is also unacceptable.

Any large system has a risk that a single problem can cause a cascading effect on its functionality or availability, as has been the case with Amazon and Google among others. One of the best ways to avoid this cascade effect is to ensure that some of your applications or portions of each application run on different platforms. Having different platforms helps to ensure that problems propagating in any one cloud platform won’t affect applications in the other.

What apps need a hybrid cloud?

As I’ve said before there is no one size fits all cloud or IT strategy. Each application or service has to be evaluated against traditional risks of outage vs. loss of business. The vast majority (70 percent or more) of the applications in an enterprise IT org don’t require anything approaching 100 percent uptime. Of the 30 percent that are critical, many will likely support the business just fine with a 99.95 percent uptime design. What you have left is the 5-15 percent of applications that the business really considers “mission critical”.

It’s also true that in some cases just 5 percent of applications could account for 50 percent or more of your company’s IT work output. So the number of applications is less important that their scope and importance. This differentiation of applications is critical to help IT teams make the decision about what applications need the resiliency of more than one cloud platform.

High reliability doesn’t have to cost double the infrastructure and people costs like you might expect from historical efforts in legacy IT. The ability to provide redundant environments or splitting your applications between two environments is a real option today when using cloud. That’s not to say there won’t be extra costs, but the costs are more related to startup investment than they are to on-going hardware and environment management and replacement costs.

In other words, if your environment is big enough or critical enough spending some upfront effort and cash to enable a hybrid cloud environment (I.e., private & public cloud or two different public environments) can’t easily be justified. In fact, there’s an online gaming company that is using the “any cloud” approach as part of their business model. The use of multiple clouds allows it to grow quickly where and when it needs to while supporting latency and availability requirements.

Complexity you control versus complexity you can’t

Once you’ve taken the steps necessary for adding another cloud provider and made multi-cloud work for critical app, it’s now much less of a hurdle to make them work for less critical environments. I also want to make clear that you can distribute one application across several clouds or split multiple applications across several clouds. The result of the multiple applications on several different clouds is lower reliability for any one application, but higher reliability for your applications as a whole. In other words, you’ll have avoided the risk of having all your apps down at the same time.

I know it seems counterintuitive to create more complexity to help protect you from complexity, but it’s not. The complexity you’re creating can be easily managed with the appropriate up-front effort and it’s under your control. On the other hand, the complexity you accept for your critical app running on one cloud is completely out of your control.

So go forth and build sustainable, supportable, highly reliable application environments across providers, in the long run it’s the smart and safer thing to do.

Mark Thiele is executive VP of Data Center Tech at Switch, the operator of the SuperNAP data center in Las Vegas. Thiele blogs at SwitchScribe and at Data Center Pulse, where is also president and founder. He can be found on Twitter at @mthiele10.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• A near-term outlook for big data
• Migrating media applications to the private cloud: best practices for businesses
• Cloud computing infrastructure: 2012 and beyond

“We are not testing Centerton. Based on how it looks on paper it doesn’t look like it meets the needs of our workloads,” said Kirkland when I asked if Facebook was testing or using the latest Intel SoC. “But we are encouraged by Intel’s general direction and we look forward to Avoton.”

Avoton is the next generation Atom core that Intel announced at our GigaOM Structure conference last June. It will be manufactured using Intel’s 22 nanometer process, which will presumably lower the energy consumption and drive production costs lower on a per chip basis. Meanwhile, the family of systems on a chip announced today are based on Intel’s Centerton technology, consume 6 watts per chip and cost as little as $54 per chip when sold in lots of more than 100. But apparently that’s not enough for Facebook, which despite its decision not to use Intel’s wimpy cores, is a big buyer of Intel’s brawnier chips.

When Frank Frankovsky, the VP of hardware design and supply chain for Facebook spoke onstage today at the Intel launch, he hyped wimpy cores in general and spent a lot of time driving home the point that they can deliver the right mix of performance and energy consumption for a lower cost. He did however spend a lot of time focusing on those lowered costs — something Intel may not want to emphasize given worries about how this segment might affect its overall margins.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• The fourth quarter of 2012 in cleantech
• Takeaways from connected consumer’s second quarter
• Ups and downs for cleantech in Q1

Bryant also added that the SoC has already made it into 20 designs so far, and that an unnamed storage vendor even swapped out an ARM design for the Intel SoC. And that’s what this press conference was all about: “Today there is no enterprise-class ARM-based server chip,” said Bryant in response to a question, but that competition is coming next year and Intel can’t afford to get caught flat footed.

After a few years of actively denigrating the idea of lower performance but power efficient chips in the data center, in 2011 Intel finally got the microserver religion, releasing a product roadmap and predicting the segment might become about 10 percent of the server market. But other than make a special 64-bit part for SeaMicro, which shoves hundreds of low-power Intel Atom chips into a dense machine, Intel talked a good game but didn’t seem to be excited about the opportunity in the way that ARM or other chip vendors such as Tilera were. It even let its rival AMD walk off with SeaMicro.

As one of our regular commenters Lucien Armasu put it, Intel is facing the classic innovator’s dilemma, by which lower-cost, good-enough products are coming from the bottom of the market to cannibalize the success of its higher-dollar cores. Lucien wrote:

Anyways, the problem for Intel is that they are very reluctant to even promote Atom for micro-servers, and it shows from how they talk about it. They have a conflict of interest, because they’d rather sell the much more profitable “bigger” chips.

The reason why this is a problem for Intel is because ARM has absolutely no problem trying to sell ARM chips for servers. In fact they have all the incentive in the world to do it, while Intel has the least incentive to do it. As Clayton Christensen puts it, Intel will be “happy to concede the low-end, non-profitable (for them) market to their disruptive competitors”.

Intel was a case study in Christenson’s Innovator’s Dilemma, so its position here is ironic. But despite analyst worries that Intel will see its margins drop from the mid-60 percent range to the 50-percent range in the coming two years (in part because it will have lower-margin chips but also because its overall competition will be tougher with ARM-based cores in the market). But Intel did note in its presentation that a highly dense Atom SoC configuration in a rack would net more revenue than a rack of fewer Xeon processors (there were roughly a fifth of the number of Xeon chips in that rack configuration). It was a bit more vague though, on the margins.

“From an Intel revenue perspective, Xeon v. Atom doesn’t matter — in fact the Atom is slightly greater,” Bruyant said. ” Across our portfolio it’s still a very good margin for us and we’re absolutely fine if the Atom SoC does well for us.”

Intel brought out a data center executive from Microsoft and one from Facebook to tout the value of this Atom SoC, although it was unclear if Facebook was actually using the SoC in its environment. Frank Frankovsy of Facebook seemed unclear on the pricing and in a discussion about the potential benefits of using this SoC, he extolled the virtue of so-called “wimpy cores” for some jobs from a price performance per watt perspective and said, “Theoretically you also drop the cost of those chips.”

It’s well known in the chip world that Facebook, as a large buyer of chips, isn’t shy about testing products from alterative chip vendors such as Tilera in hopes of keeping pricing pressure on Intel. What’s more interesting is if other vendors start doing this. At that point, Intel may have the products and features on its chips that webscale customers want, but it will have to be competitive on price in a way that it really hasn’t had to be when it was just competing against AMD.

Facebook has said it will use these wimpy cores(one promising segment might be in its planned cold-storage architecture), but the jury is out on which vendor will win its business. Intel clearly intends to fight for Facebook and other webscale business, but when it does, it’s going to be facing up against some real competition — competition that it hasn’t faced before in the Intel/AMD world.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Power in the data center: Can it drive disruption?
• Infrastructure Q2: Big data and PaaS gain more momentum
• The fourth quarter of 2012 in cleantech

In the data center, the flood of information traveling across networks has changed in three ways. First, there is more of it. Second, the data is related to other servers inside the data center as opposed to getting a request for data and then serving it up. This is the so-called East-West traffic explosion. Instead of a server sending information up to a switch that sends it out of the data center, the server is now sending requests to a switch that then connects to other intra-data-center servers. Thus one request can now involve a few switches and several servers sending traffic back and forth across the network before it ever leaves the data center.

The third change is that cost pressures and the need to scale are pushing data center operators to flatten out the network so more servers (or virtual machines) talk to the switch. The solution here so far appears to be a networking fabric, but may in time graduate to a truly distributed and virtualized network.

The net result of these changes is that network pipes need to be fatter while the network processors need to be faster. But because this is also a data center, the components that enable this have to be relatively cheap. This is where Kotura comes in. The startup, based in Monterrey Park, Calif., offers a fiber-based transceiver that can deliver 100 gigabits per second inside the data center. The transceiver could live on a board next to the CPU or inside a switch and could eventually expand to deliver a terabit per second (Tbps).

While one 1 Tbps is crazy fast when you consider that many data centers are currently upgrading to 10 gigabit Ethernet between servers, it’s going to be necessary. Arlon Martin, VP of Marketing, Government Contracts & Industry Relations at Kotura, tells me that customers are building products for the high-performance computing sectors but also for real-time data processing. The goal is bringing a low-power and less expensive optical part into a rack of servers, able to scale up to terabit per second capacities.

If you have a rack of servers with even multiple 10 GigE ports, suddenly the top-of-rack switch, or whatever fabric is stitching those servers together needs to have a lot of bandwidth. This is something Cisco has noticed, which is why it purchased Lightwire earlier this year. Kotura provides it, and because its chip stuffs 25 Gbps into a single wavelength in a strand of fiber with up to 40 available wavelengths customers can light up the remaining wavelengths as needed to reach up to a terabit of capacity.

Some other silicon photonics vendors require more chips or line cards to upgrade, which means new equipment. Much like Plexxi, which is bringing fiber-based gear to the data center, Kotura is betting that scaled-out networks have a need for speed (and capacity) that only fiber can provide. However, Plexxi wouldn’t compete with Kotura, but would likely buy a chip from it (or another transceiver vendor) to power its gear.

Kotura has 75 employees and has raised undisclosed millions from ARCH Venture Partners, Fuse Capital, GF Private Equity and others. It has an established customer base in the telecommunications business where it has sold product since 2006. But now it’s moving into the data center in the hopes of solving looming problems in the networking sector with cheaper, low-power optical chips that can deliver a lot of capacity between servers. The new data center is going to have a lot of dense computing, low-cost fast storage, and soon, high-capacity low-latency networks connecting everything.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Cloud computing infrastructure: 2012 and beyond
• What Enterprise Software Vendors Could Learn from the Consumer Space
• The promise of SDNs in the enterprise

From the design challenge web site:

Servers are typically replaced about every four years. This is necessary to maintain fast, reliable equipment. Unfortunately, this results in a lot of waste. Open Compute wants to change this starting with the server chassis. These are typically made of steel, which is recyclable, but even recycling generates waste. What would happen if these chassis could be placed in compost instead?

Purdue University’s College of Technology entrepreneurship program, called Tech Ventures, will work with the Foundation and the social network to get students to rethink the humble chassis. The challenge will begin with a Computer and Information Technology (CNIT) course at Purdue in the spring semester. I, for one, can’t wait to see the cardboard server chassis, although in some ways I think eliminating the chassis and making the rack the protective unit for the boards and components might make more sense.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Locating data centers in an energy-constrained world
• A near-term outlook for big data
• Webscale and cloud are changing the server value chain. Who wins?

You may not get the budget or buy-in to create a real business continuity plan, but if you can’t get one, then a communication plan might just be the best return on investment for the time and cost.

It seems simple, but communications is the single most critical capability post disaster. If your customers don’t hear from you, they assume you’re either out of business, or at a minimum unlikely to keep near term commitments. As anyone calling a customer support center will tell you, the black hole of “not knowing” is worse than being told, “it might be another day.” With a timeline the customer can react and plan, without it, she is at the whim of her vendor, which is now making her look bad with her customers.

Setting up the plan

Lobby at Verizon office at 140 West Street, New York post-Sandy

• Have a phone line with regularly updated information for all employees, with a tree for updates, or information that might be critical to sub groups.
• Create a website that is used only for updates.Be sure to host your recovery site with a hoster who is out of your area.
• Get a conference line(s) that are available for teams to jump on whenever they need to discuss specific activities with other members of the recovery team.
• Provide a location for meetings with at least two backup alternatives in case the primary isn’t an option. This could be someone’s home, some rental space or a conference room at a local hotel.
• Distribute pagers and or satellite phones for a few key staff. This can be especially important if the cell systems have been impacted by the disaster.

The tools of the plan

Email: Create a mirror of your email environment with a third party (if you haven’t already outsourced it to Google or Microsoft, etc.). This mirror doesn’t have to have all the data, it just has to have key names, addresses, and distribution lists. Your email should appear to be coming from the same domain as your primary email, and it should have the following:

• Mirror email accounts for a team of leaders and functional team members that are tasked with responding after a disaster;
• Addresses for all your customer contacts;
• Addresses for your key suppliers; and
• Distribution lists for customers and employees.

It really isn’t hard to make the above happen. Your email admin can find a third-party provider and setup a mirror but inactive environment with regular synchronization of the required information. This mirror environment allows you at low cost to have a “known and trusted” communications tool up and active the minute your primary systems are down.

Phones: Like the email solution, you should have a backup voice service such as Centrex. This voice service will have all the same contact information that your email solution should have, plus more depending on special contact details for recovery teams and key suppliers, etc.

Social media: If you’ve been reading the news lately, you‘ve seen or heard information on the importance of your CEO participating in social media (Twitter, Google+, Facebook, Quora, etc). A website with blogs and or a dashboard with updates can also be very useful. After a disaster your ability to communicate updates and answer questions in real time from customers, analysts, partners etc. is a huge benefit. The more information you make available the more comfortable your community will be and the more patience they will show you during the recovery period.

Justifying the communications efforts

During the disaster and the recovery these efforts will let you tell the customer know what’s going on. You can provide comfort to concerned families, partners and employees. Simple messages indicating what you’re doing, what your update schedule will be and how company representatives can be reached are all key objectives of the communications effort.

Of course, each communication tool has its advantages and disadvantages. I don’t recommend social media as the primary communication solution in lieu of email and phones. You still need the ability to provide targeted information to your customers and partners that would go way beyond the hit and miss capabilities of social media. Each of the tools listed above have capabilities that make them more suited for certain types of communication requirements.

In the case of social media the updates are really general and meant to satisfy the casual observer, while also updating those who do follow your tweets, LinkedIn updates, and Facebook posts closely. However, you shouldn’t consider it a substitute for direct communication with customers/partners via phone or email.

While these efforts aren’t an alternative to a real business continuity effort, like a weight loss plan or cleaning out the garage, taking small but measurable steps is often the key to making “something” happen. Each of the objectives listed above are relatively inexpensive and very easy to explain or justify to your leadership. At a minimum it will demonstrate that you’re taking the lead on “doing something” rather than ignoring the issue and blaming corporate inertia. You’ll certainly be doing your company a huge favor by ensuring that they don’t appear to disappear from the face of the earth when a disaster occurs.

Mark Thiele is executive VP of Data Center Tech at Switch, the operator of the SuperNAP data center in Las Vegas. Thiele blogs at SwitchScribe and at Data Center Pulse, where is also president and founder. .He can be found on Twitter at @mthiele10.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• NewNet Q2: Google closes the quarter with a bang
• Finding the Value in Social Media Data
• What Enterprise Software Vendors Could Learn from the Consumer Space

With the PC market dwindling, AMD has little choice but to look elsewhere, and servers might be its saving grace. While it’s a tough overall economic environment for chipmakers, Intel’s third quarter financial results from last week show that the data center is at least still a growing market. Intel reported that its PC revenue was $8.6 billion, down 8 percent year-over-year while its data center group revenue was $2.7 billion up 6 percent year-over-year. Yes, that’s a smaller base compared with PCs, but it’s still growing.

On Monday afternoon, AMD sent out an email inviting reporters to an event featuring CEO Rory Read, Lisa Su, senior vice president and head of AMD’s global business units, and a “special guest.” Given that ARM is hosting is developer conference in Santa Clara, Calif. that week, and that AMD’s invite notes that this meeting will involve discussion of its “ambidextrous strategy,” I’m hoping that it is finally biting the bullet and taking a higher-level ARM license that will give it the ability to design chips for servers — something I predicted it would do back in February when it purchased SeaMicro.

AMD’s big, fat strategic shift

AMD has already started tiptoeing down this path with the creation of the Heterogeneous Systems Architecture Foundation back in June. It joined up with ARM to create an organization that will bring heterogeneous computing — i.e., computing that doesn’t just use x86-based chips — to the mainstream. It also has licensed a lower-end ARM-based core for use in security. So taking it further with the license of a Cortex A15, or higher-end server-capable chip, isn’t going to shock investors at this point. It may even hearten them.

AMD data by YCharts

AMD has a history of innovation that has been stymied by Intel and its own execution troubles; sometimes both. AMD brought 64-bit computing to the x86-based architecture in desktops and servers, introduced its HyperTransport technology to improve scaling of multiple cores ahead of Intel and even foresaw the importance of graphics chips for both consumer and corporate computing back in 2006 when it purchased ATI. Unfortunately it also failed to go big in mobile. AMD did purchase Alchemy, a low-power MIPS-based chip company back in 2002 as part of mobile bid, but in 2006, it got out of that business.

In a similar vein, Intel has its own ARM-based Xscale unit during that time that it later sold to Marvel. But even after the launch of the iPhone and netbooks (remember netbooks?) AMD stuck to its plan to steer clear of mobile computing. And that was a huge mistake. AMD recently said it would lay off about 15 percent of its staff, and also reported lower-than-expected sales.

Some of the blame can be laid on AMD for not embracing mobile, but there’s a larger upheaval occurring in the chip sector where industry segments that were once clearly delineated have begun to merge (like smartphones and computers) while the hegemony of the x86 architecture with its two primary competitors has waned. So there’s competition between more players and more architectures across a wider spectrum of devices. It’s happening clearly in consumer devices, but it is about to happen in servers.

The ARM servers are coming!

Calxeda’s ARM-based server chip.

At this point, AMD has — with its SeaMicro purchase — systems design expertise and IP to use an ARM-based chip inside a server built for cloud computing and big data jobs. It has credible x86-based processors as well as GPU-based systems designed for high performance computing. Bringing in an ARM license and experienced ARM engineers means AMD could build machines that mix and match the most popular architectures for different styles of computing. Yes, it will compete with many more players than Intel: already Marvell, Cavium, Calxeda and other chip companies are targeting specialized chips for the server market.

Just last week Calxeda launched a roadmap that brings 64-bit ARM-based computing to servers in 2014, and says it will have ARM servers in production environments by the end of the year, while other vendors such as HP and Dell are launching ARM-based servers. I expect more news at the ARM conference next week, but the momentum around the technology is clear. After years of dithering, it’s time for AMD to bet it all on heterogeneous compute, even when it comes to its server business. Will it be enough?

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• The fourth quarter of 2012 in cleantech
• Cloud computing’s impact on chip and hardware design
• Cloud computing infrastructure: 2012 and beyond

As of today, it has a powerful top of rack switch that will doubtless cost less than Cisco’s gear, and it has a distribution deal with Synnex, a huge contract manufacturers and reseller of IT gear. Basically, Huawei has launched the goods to fight Cisco (and Juniper) in the enterprise networking market and has the means to get it in the hands of customers. Huawei has shaken things up in the carrier equipment market for quite some time, precipitating consolidation and lower pricing by major vendors, as well as a rush into services. And now that change is about to break like a wave in the data center world.

However, unlike the carrier gear market, Huawei isn’t entering a staid business as a newcomer with low prices and the rumored support of the chinese government (and financing arm). Enterprise networking is a far more competitive world with a variety of companies pushing into the unified communications and video conferencing market at both the high and low end. Cisco’s at the high end with its Telepresence suites, and Skype and Logitech are at the lower end. Huawei’s bringing its high-end telepresence product to market to compete directly with Cisco’s high end telepresence products.

In the switching market, Huawei is launching a top of rack switch and a series of core networking switches that seem aimed at Cisco’s Nexus line of boxes. Huawei’s CloudEngine 12800 series of switches provide up to 48 Tbps of capacity and bandwidth of 2 Tbps per slot which is more than the current Cisco gear in both cases. Aside from the threat to Cisco, the data center networking market is undergoing several shifts, some of which will play to Huawei’s advantage.

There is the growing popularity of whitebox switch makers using merchant silicon as opposed to proprietary chips, which webscale and cloud data center operators are buying en masses according to Arista and the CEO of Pica8, another vendor aiming to provide whitebox networking gear. The scale-out players are indicative of another trend, which is the increasing scale and changes in traffic patterns among data centers. Data centers are larger and the networking is growing more complex. Thus vendors are demanding lower cost gear, such as that provides by whitebox vendors, but also eyeing the creation of software defined networks.

Thanks to these trends, most people in the networking industry expect to see huge changes taking place in the next five years. As a new player with a lower cost basis, Huawei is poised to take advantage of these trends. No wonder Chambers’ is worried.

Image courtesy of Huawei.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• The promise of SDNs in the enterprise
• The promise of software-defined networking
• Cloud computing infrastructure: 2012 and beyond

Facebook’s Prineville data center.

Building sustainable data centers is hard — especially if you’re trying to do it in office space in Houston. Plus, the idea of operating some kind of power-generation plant for offering renewable energy such as solar or biogas is a scary prospect for data center operators. These were among the key takeaways (along with a few less-obvious lessons) from a panel on sustainable data centers at the Open Compute Summit held today in San Antonio, Texas.

Bill Weihl, manager of energy efficiency and sustainability at Facebook and the former energy czar at Google, moderated the panel that also featured Melissa Gray, the head of sustainability for Rackspace; Stefan Garrard who is building an HPC cluster for oil company BP; Winston Saunders from Intel; and Jonathan Koomey, a consultant and energy-efficiency expert. While we are entering the age of 100-megawatt data centers the size of football fields, we’re also dealing with higher energy costs and concerns about how to keep our webscale infrastructure running. As part of the focus on lowering costs, The Open Compute Project spends a lot of time on sustainability.

But lowering the energy inside a data center can only go so far. Saunders explained that chips for example, had achieved their lowest possible power utilization without new breakthroughs. Even when idle, the chips still consumer 20 percent of their maximum energy draw because they can’t fully turn themselves off. The inability to power all the way down is a function of adding latency (once something is turned off it takes time to turn it back on) and because powering down the chip requires the data center to stop sending information. However, data centers rarely hit that point, which means chips are always “awake” and consuming energy.

But it’s not just the hardware. Garrard said his current high-performance computing cluster is running in office space that holds both humans ans servers. He’s done a little to help make things more efficient, but because of the office location and Houston’s hot and humid climate, his servers run at a power usage effectiveness of more than 2 (Facebook, which has heavily optimized its PUE is about 1.07; 1.0 is ideal). So, he is building out a new facility and hopes to get closer to a PUE of 1.5.

But where will the power for his and other new data centers come from? Renewables aren’t really on the list yet. When asked about using biogas systems such as those from Bloom Energy or solar, Gray said the idea of running a generation plant along with a data center was so far outside her core competency that it wasn’t really something she thought about.

Koomey, however, called the idea that a data center operator has to follow in Apple’s footsteps to operate their own generation (Apple is using Bloom’s boxes to power part of its new data center) a “canard” and said data center operators should get renewable power from their utilities. Weihl, who helped Google buy wind power from providers for its data centers, agreed.

The panel essentially outlined several areas where data center infrastructure consumes energy. In the ideal world, operators could site their data centers in places that are cool and dry, and build out the ideal facility and hardware to reduce the power draw. As Koomey said, they could think “holistically.”

Unfortunately, most data centers are built in the real world, where and when they are needed with the equipment available at the time. The standards and designs offered by the Open Compute Project will help, but the real world will take its toll.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How tomorrow’s mobile-centric data centers will look
• Takeaways from the second quarter in cloud and data
• Cloud computing infrastructure: 2012 and beyond