I spent a day last week at CTIA Wireless in New Orleans, having some productive discussions with a range of companies. While this show certainly seems to have steadily become much more consumer-oriented, there were still a number of interesting announcements and exhibits from the infrastructure side of the business.

The overriding theme of the show seemed to be the growth in LTE deployments, subscribers and devices. At the time of last year’s show in Orlando, most of the 4G devices available in the US were based on either HSPA+ or WiMAX and the first LTE smartphone had just been introduced. One year later, multiple LTE smartphones are available while Sprint and Verizon are rapidly deploying LTE coverage in the US.

A lot of interest at the show, in fact, was generated by the on-going debate between Sprint and Verizon about the performance of their respective networks. During a keynote address, Verizon’s CEO Dan Mead suggested that Verizon’s network would deliver higher performance thanks to its use of 20MHz of spectrum, whereas Sprint’s is restricted to 10MHz. Sprint’s Bob Azzi subsequently stated during a media briefing that, while their narrower band would cause a difference in peak speed, the overall subscriber experience would be “very good”, implying no discernible impact to overall performance. Sprint also discussed plans to use Clearwire’s TD-LTE service to supplement the capacity of its network and then move to LTE-Advanced in the future.

Small cells were also a hot topic at the show. While the small cell concept is perceived as the optimum approach for operators to deal with booming mobile data demand, there are now signs that it may be hard to derive the expected benefits from them in some cases. In theory, small cells make more efficient use of existing frequencies and cover areas, such as indoor spaces, that are hard to reach with macro cells. However, there are significant challenges in locating and configuring the necessary radios: since standards are still in flux, there may be hidden costs behind the relatively low prices and carriers will likely end up competing over choice locations. To prevent interference, macro cells and small cells need to be coordinated and, while the Small Cell Forum has advocated common, open standards for this, there is a risk that vendor-specific protocols will be the de facto scenario. Other issues are more practical, such as the concern that even though the small cells cost much less than macro equipment, each still needs to have a fast backhaul connection, which is usually wired. Small cells mean more cells in a given area, implying more wires and bandwidth charges for the mobile carrier.

Given 6WIND’s focus on delivering solutions for Software Defined Networking (SDN), I was pleased to have a long chat with a couple of executives in Nokia-Siemens Networks’ booth about their Liquid Net initiative. From a branding perspective, the name is an ideal summary of the concept whereby network resources “flow” to where they are most needed, minimizing unused resources and maximizing overall resource utilization. The “Liquid Radio” solution applies Cloud RAN principles to baseband pooling of traffic from active antennas, while the “Liquid Core” virtualizes core network applications to maximize the hardware efficiency and flexibility of the core network.

I didn’t get to spend as much time at the show as I would have liked, so I’d be very pleased to hear your views. What were the key trends that you observed? What product announcements were most interesting?

I spent a day recently at InfoSecurity in London (http://www.infosec.co.uk/) which with over 12,000 visitors is one of Europe’s largest security conferences.

I talked with many exhibitors about network security along with their challenges in migrating from proprietary environments to more flexible platforms, based on x86 processors and with virtualized architectures. It seems that the security market is working hard to bring more agility to network security products while still offering advanced security functions. Indeed, almost everyone I talked to agreed that Software-Defined Networking (SDN) solutions, increasingly requested by service providers and data center operators, are key to delivering the SLAs that end-users expect, while ensuring acceptable ROI for the service providers themselves.

It is not surprising the difficulty of predicting application trends, along with the boom in sales of smartphones and tablets, present major challenges to service providers who must predict and adjust finely their capacity while keeping costs at a reasonable level. This is a key driver for the deployment of SDN technologies in order to deliver the same benefits for networks that server virtualization already provides for compute subsystems.

I talked to suppliers of next-generation firewalls, Unified Thread Management systems, Intrusion Prevention Systems, Application Delivery Controllers and WAN Acceleration Controllers, who confirmed their need to re-architect their solutions to x86 new Intel architectures while maintaining the performance levels reached by proprietary hardware. Additionally they are all exploring the adoption of virtual network appliances.

On another topic, I talked to various OEMs working on plans for improving their DPI engines, since they see a convergence between Layer 4 through Layer 7 security services. Policy enforcement, cyber security, URL parsing and deep layer 7 protocol analysis become a requirement for many of their products. But there are significant challenges in the way of implementing these services on standard servers, such as how to terminate TCP or SSL sessions for Layer 7 DPI, how to implement crypto engines in a virtualized environment and how to architect a smart “flow table” for triggering DPI engines.

How do you see market trends in the security domain? Are you working on some form of security convergence? What are your plans for hypervisor optimization and how would you implement a smart flow table on an x86 platform?

I spent a day this week at Interop in Las Vegas. Interop is always a fascinating show because it covers the whole range of products and technologies that can loosely be termed as “networking”. The exhibits span everything from cables and server enclosures to high-end networking equipment and a wealth of cloud-related software.

Given 6WIND’s focus on Software Defined Networking (SDN), it was interesting to see a significant focus on this topic, as well as on OpenFlow which is unquestionably perceived as an unstoppable trend bringing major advantages in terms of network manageability, scalability and testability. I particularly enjoyed a multi-vendor presentation in NEC’s booth, describing their ProgrammableFlow solutions, which featured speakers from Brocade, Extreme Networks, IBM, Intel, Microsoft, NEC and Radware. It was also interesting to see a multi-vendor interoperability demonstration at the OpenFlow Lab, where OpenFlow controllers and switches from companies like Big Switch, Brocade, Broadcom, Citrix, Extreme Networks, Huawei, HP, IBM, Intel, Ixia, Lyatiss, On-Lab, NEC, NETGEAR, Netscout, Spirent  and Vello Systems appeared to be working together.

With all the industry hype around SDN and OpenFlow, it seems that Cisco has been relatively quiet about their plans recently. So the opening keynote address by Padmasree Warrior, Cisco’s CTO, was a welcome (and well-attended) kickoff for the show.

Warrior began her talk by summarizing her view of the four-stage evolution of networking technology. Following basic connectivity (browser, e-mail, and search) and then the “digital revolution” highlighted by e-commerce and IP telephony, she positioned the current stage as the “networked economy” (social media, mobility and cloud) and described the future as being a “human network” with immersive data and the “Internet of Things”.

Warrior showed a study indicating that by the end of this year 70% of enterprises will use cloud technology at some level. She mentioned that 56% of companies want desktop virtualization and discussed statistics demonstrating that a cloud-based data center with 1,000 servers or more has a Total Cost of Ownership (TCO) 50% less than a traditional data center.

A key projection in Warrior’s presentation was that the Internet will double in size every five years. She emphasized that video is, and will remain, the biggest issue for service providers, indicating that video traffic will quadruple between now and the end of 2014, while two-thirds of overall Internet traffic will be video by the end of 2015. She stressed that “the number one priority for CIOs is video” and that the convergence of cloud and mobility is driving new, collaborative work styles that involve massive real-time video traffic.

Warrior’s vision of the new “intelligent network” is one that’s designed around six key features: it’s visible, aware (e.g. real-time analytics), secure, programmable, agile and manageable.

So what is Cisco’s strategy in this area? Warrior stressed Cisco’s commitment to the concept of SDN, mentioning the recent “spin-in” acquisition of Insieme, and discussed Cisco’s use of OpenFlow to partition the network between the control plane and data plane while providing access and programmability at multiple levels yers within the overall stack. She talked about Cisco’s “Open Programmable Environment” architecture which includes APIs at multiple layers in addition to the control plane and data plane, though she didn’t provide much detail on this (many in the audience seemed to want to know more).

Warrior summarized by describing the four principles that Cisco is applying to next-generation networking: customized (tailored) solutions, open ecosystems, partnering and innovation.

Were you at Interop this week? What were the key trends that you observed? What product announcements did you see that are most relevant to the topic of multicore packet processing? What did you see as the highlights of this Cisco keynote address?

In a recent post, Charlie summarized his discussions at the Open Networking Summit. The conference covered a wide range of business and technical subjects relating to Software Defined Networks (SDN), with the topic of OpenFlow receiving particular attention in the networking industry.

Last month I visited Japan and made a presentation about 6WIND’s technology to a group of Japanese journalists. Japan is one of the countries with the highest penetration of LTE worldwide and a major market for 6WIND as the 6WINDGate software is now serving 80% of Japan’s LTE users. Japan is also viewed as the leader in deploying OpenFlow and SDN. So, I was asked a lot of questions about the two following topics:

• How does 6WINDGate fit the OpenFlow concept?
• How will SDN help the convergence between mobile and cloud infrastructure?

Regarding the first question, the 6WINDGate architecture was designed 6 years ago based on the same concept as OpenFlow: the separation between the control plane and the data plane. OpenFlow defines the architecture and a protocol between Controllers and Switches to simplify network configuration and operation. The separation between the control plane and the data plane also enables a high performance packet processing architecture for the data plane, which is the focus for 6WINDGate.

Today, OpenFlow mainly addresses protocols for Layer 2 switches. However networks need more sophisticated services, for which 6WINDGate brings added value: high performance and enhanced data plane features to provide rich networking services such as virtual routing, encapsulation, firewall, IPsec, MPLS, etc. 6WINDGate is available today to support future versions of the OpenFlow specification that will address these enhanced services.

This leads to the second question. The new generation of mobile infrastructure (LTE) relies on fully software-defined solutions like 6WINDGate to implement complex networking features and meet requirements for high performance, flexibility and scalability.

These solutions can also address challenges in data centers, such as total cost of ownership as well as performance and flexibility. They deliver high performance on commodity server platforms, while accelerating the use of virtual appliances and enabling the use of Network-as-a-Service solutions for improved monetization.

The mobile infrastructure market is now integrating cloud concepts by using generic platforms to implement multiple network software functions, bringing tremendous cost savings and greater flexibility.

High performance packet processing implementing complex network services and removing network performance bottlenecks in virtualized architectures is a key technology in ensuring a true convergence between mobile and cloud infrastructure.

I spent a couple of days last week at the Open Networking Summit in Santa Clara. The conference covered a wide range of business and technical subjects relating to Software Defined Networks, with the topic of OpenFlow receiving particular attention.

In a keynote address on the first day of the conference, Urs Hoelzle, Senior Vice President at Google, described how Google used OpenFlow as part of a major overhaul of its internal networking infrastructure to reduce costs and increase efficiency. This internal “G-scale” network connecting Google’s vast data centers worldwide actually carries more traffic than its external “I-scale” customer-facing network.

Hoelzle presented a fascinating description of how Google approached the massive task of swapping their network hardware while, of course, ensuring no disruption to on-going traffic. Having concluded when they started the project that there was no off-the-shelf networking equipment “even remotely suitable for this task”, Google developed their own 128x10G chassis using merchant silicon, ready for deployment starting in early 2010. They then installed this in each data center, using a method whereby they first pre-deployed the equipment at a site, taking down half the site’s networking equipment and connecting it to the new system. After testing to verify that the upgrade worked, the engineers would then repeat the process for the remaining 50 percent of the site’s equipment. By early this year, Google’s entire internal network was running the new, OpenFlow-based equipment.

Google now has a central traffic engineering controller that optimizes routing and provides more predictable performance. OpenFlow allows network packets to be routed by software running on routers. By separating packet switching from network management, Google has better and easier control over their network.

Many speakers pointed to Google’s announcement as a key proof point for OpenFlow, demonstrating that the technology is usable in the world’s largest networks and delivers quantifiable business benefits. This view was certainly reinforced by other case studies presented at the conference, by organizations as diverse as Kindred Healthcare, Indiana University and NTT Communications.

One of the questions debated at the conference was how quickly OpenFlow will gain traction in broader enterprise and data center applications. A couple of speakers pointed out that Google of course has access to resources way beyond what’s available to most IT organizations, while service providers are naturally hesitant to introduce new technology into something as complex as the average enterprise network.

Martin Casado, Nicira’s CTO, talked about the value of OpenFlow within virtual switches (announcing that the OpenFlow Virtual Switch is now included in the Linux kernel), but recommended using Layer 3 to manage the external fabric interconnect. Speakers from Dell, HP and other companies mentioned that merchant silicon (e.g. switches) is not yet available with fully optimizations for OpenFlow.

A common thread within the presentations was the likely coexistence between existing networks and SDN. Speakers from Cisco and Juniper both talked about this, explaining that they expect to leverage traditional routing protocols with SDNs for the foreseeable future. Cisco described how SDNs should be linked to analytics and policies, using visibility into the control, data, management and transport planes to drive more intelligent decisions.

Finally, there were some interesting discussions about what “killer apps” might emerge as the driving force behind Software Defined Networking and OpenFlow. Network virtualization was mentioned, as was fine-grained policy for Bring Your Own Device scenarios. A couple of people talked about a future with “app stores” providing off-the-shelf networking applications that would (presumably) run on standard, secure, risk-free networking APIs. Personally, I would settle for not having to reboot my home router every time there’s a full moon, or a decent GUI for updating its firmware.

What are your thoughts on this topic? What do you expect to happen in terms of OpenFlow’s adoption in enterprise networks and data centers? What are the key issues (both business and technical) that the industry needs to resolve to accelerate this adoption? Is there a “killer app” for this technology that will trigger a sudden increase in deployments?

By Charlie Ashton, VP of Marketing – 6WIND

The event that used to be known as Embedded Systems Conference (ESC) has now been combined with several other conferences (including the former Multicore Expo) to create DesignWest, held last week in San Jose, CA.

Being an engineer at heart, I have always found ESC to be a fascinating event in terms of the breadth of products featured in the expo. Everything that an embedded system developer could possibly want is typically on display: boards, development tools, application software, processors, DSPs, analog components, sensors etc. This year’s event was no exception, though the crowds did seem to be somewhat smaller than in 2011, maybe because it was Spring Break for many local schools last week.

In terms of product announcements relating to multicore packet processing, and therefore most relevant to readers of this blog, it was interesting that both Altera and Xilinx were heavily promoting platforms combining multiple ARM cores with an FPGA fabric.

Altera’s SoC FPGAs integrate a processor subsystem (comprising a dual-core ARM® Cortex™-A9 MPCore™ processor, a rich set of peripherals, and a multiport memory controller) together with a flexible FPGA fabric and a high-bandwidth interconnect. Hardened embedded peripherals in the processor subsystem eliminate the need to implement these functions in programmable logic, leaving more FPGA resources for application-specific custom logic and reducing power consumption. At the same time, the flexibility offered by the FPGA logic fabric lets developers differentiate their systems by implementing custom or off-the-shelf preconfigured IP into their designs.

In the case of Xilinx, the Zynq-7000 Extensible Processing Platform (EPP) was named SoC Product of 2011 in the UBM Electronics ACE awards for Ultimate Products of the Year. The Xilinx Zynq-7000 EPP is a family of devices that combine a complete ARM dual-core Cortex-A9 MPCore processor-based SoC with integrated programmable logic. Each device is a processor-based system, capable of booting an OS from reset without needing to program the programmable logic. It enables developers to apply a combination of serial (using the ARM processor) and parallel (using programmable logic) processing to applications that require high levels of performance.

It’s interesting to consider how these kinds of products might be used in high-performance packet processing systems. The control plane would obviously run on the ARM cores and it would seem that certain performance-critical functions would be implemented in the FPGA fabric for maximum performance, analogous to the on-chip accelerators on conventional multicore SoCs. This would enable the data plane, running primarily on the ARM cores, to leverage those FPGA-based functions to maximize overall system performance.

The opportunity for OEMs, of course, is to differentiate their products by implementing proprietary, application-specific networking functions within the FPGA fabric. However this needs to be accomplished in a way that retains compatibility both with the application software and with the packet processing stacks, while also enabling the right level of debugging and profiling.

What are your thoughts on how these CPU-plus-FPGA solutions will be used in high-performance packet processing applications? What are the software architectures that will make the most sense?

FierceWireless recently published an interesting analysis of how new “freemium” 4G data services might impact major carriers worldwide, prompted by NetZero’s launch of a completely free tier of wireless data service. After buying a $49 USB modem, NetZero users can get 200 MB of Clearwire’s WiMAX service per month for free. And other providers such as FreedomPop have indicated plans for even more aggressive packages. The article explores how the use of these freemium services can result in significant cost savings for individual subscribers and analyzes the potential loss in revenue for carriers.

FierceBroadbandWireless reported that the worldwide LTE subscriber base nearly doubled in Q4 2011, reaching 12 million users with quarter-on-quarter growth of 92%, in contrast to WiMAX which grew its subscriber base by only 14 percent in the same period. 54 operators worldwide had launched LTE commercially by the end of 2011, with the report predicting that 469 million LTE subscribers will be active by 2016. The leading equipment suppliers during Q4 were the combined Nokia Siemens Networks and Motorola (26% of LTE contracts), Huawei (23%) and Alcatel-Lucent (12%).

Orange/France Telecom confirmed its aggressive LTE plans. As reported by telecoms.com, the carrier has committed itself to rolling out LTE networks across ten European countries by 2015: Armenia, Belgium, France, Luxemburg, Moldova, Poland, Romania, Slovakia, Spain and the UK. The move is designed to help the EU reach targets set out in its Digital Agenda, which aims to ensure that everyone in Europe has access to broadband by 2013, assisted by the release of a harmonized band of radio spectrum for use with LTE.

In a report on the cloud computing space, Infonetics released a study showing that overall sales of data center networking equipment market (Ethernet switches, application delivery controllers and WAN optimization appliances) grew by 9% in 2011, with double-digit growth forecasted for 2012 and 2013. Cisco continues to be the market share leader in this segment due to the strength of its data center Ethernet switch business, with F5 and HP jointly in second place.

What are the industry trends that you’re seeing that have implications for multicore platforms? What recent interesting news have you seen in this area?

In a joint press release last week (click here), 6WIND and TI announced that multicore processors based on TI’s recently-announced KeyStone II architecture will be supported within the 6WINDGate™ packet processing software, with this support becoming available in the second half of 2012.

TI’s scalable KeyStone II architecture includes support for both TMS320C66x digital signal processors (DSP) generation cores and multiple cache-coherent quad ARM® Cortex™-A15 clusters, for a mixture of up to 32 DSP and RISC cores. The multicore architecture includes capacity expansion for SoC structural elements such as TeraNet, Multicore Navigator and Multicore Shared Memory Controller (MSMC). This expansion allows developers to fully utilize the capability of all processing elements, including ARM RISC cores, DSP cores and enhanced AccelerationPacs. RISC processing within KeyStone II has been significantly upgraded with the addition of quad ARM Cortex -A15 clusters, providing ultra-high performance at half the power consumption of traditional RISC cores.

This announcement of KeyStone II support means that 6WINDGate is now available on a total of four multicore CPU architectures. Following initial releases on MIPS architecture processors from both Cavium and RMI (subsequently NetLogic and now Broadcom), support has been extended to include Intel® architecture-based platforms, to the TILEPro64 family from Tilera and now to ARM-based processors from TI.

Our recent experience certainly indicates increased diversity in multicore CPU architectures chosen by OEMs for use in high-performance packet processing. I wonder how long this will continue, and whether the industry will consolidate to one or two architectures in the future: maybe one that is available as licensable IP and one that is only available in SoCs?

What do you think? Are the requirements for high-performance packet processing applications varied enough, and the unit volumes large enough, that this architectural diversity will continue? Or will there be a consolidation at some point and, if so, what factors will determine the winner(s)?

We’d be very interested in your thoughts on this question. And regardless of the directions that the industry takes in the future, 6WIND plans to continue to support the multicore architectures that our customers select, enabling them to bring best-in-class solutions to their end-users.

In last week’s blog I mentioned a few of the interesting new products featured at RSA Conference and discussed one of the common threads from conversations with attendees, which was the need for increased performance and scalability in the low-level packet processing functions that are a critical part of many of these products.

From a system architecture perspective, we also heard consistent comments about the need for efficient implementations of hybrid systems based on multicore architectures.

In this context the term “hybrid system” refers to one comprising multiple (typically, two) processor architectures. Generally, one architecture (generally x86) is used for the control plane and another for the data plane. The data plane architecture can be either a standard multicore platform such as Cavium or NetLogic/Broadcom, or in some cases a specialized Network Processor from a company such as Netronome.

It’s interesting to talk to system designers about the tradeoffs that they consider when choosing a hybrid system architecture as opposed to a unified approach (where both the control plane and data plane are implemented on the same architecture and, often, on the same processor).

One the one hand, a unified system brings compelling advantages in terms of a single software development environment, a more straightforward hardware design and the ability to work with a single processor supplier.

A hybrid system approach creates the challenges of dealing with two (generally very different) software development environments, two Operating Systems, software integration issues and a more complex hardware design. Also, the overall project schedule is now subject to the delivery dates for two multicore processors rather than one.

The message that we received from attendees at RSA Conference is that, in specific cases, the increased complexity and risk of a hybrid design is still outweighed by the overall system-level advantages that can be achieved. Generally, these advantages are reduced system cost, higher overall system throughput and increased performance on specific security-related functions which typically are performed in dedicated offload engines in a hybrid system.

Within 6WIND’s software, we have recognized this need and provide highly-efficient support for hybrid system architectures. All the 6WINDGate control plane code and most of our data plane code is architecture-independent. Architecture-specific abstractions and optimizations are performed within a data plane module called the “Fast Path Networking SDK” or “FPN-SDK”.

From our customers’ point of view, they license from us the 6WINDGate stack including the protocols that they need, along with the appropriate FPN-SDK modules for the processors that they select. In terms of our development process, this architecture enables us to work on both new protocols and new architecture support as parallel activities, bringing a high degree of efficiency to our new product development process.

What do you see as the trends and tradeoffs in terms of hybrid vs. unified multicore systems in networking and security applications? Will there always be a market segment where hybrid systems are the optimum solution?

Please find hereafter a selection of press releases and announcements at Mobile World Congress 2012.

Processors:

LSI announced the addition of ARM’s latest multicore technology to expand the Axxia Platform (Press Release). This is an interesting move as it seems more and processor vendors are now giving up the PowerPC architecture and choosing ARM for their future generations of products

III will be the key number for Cavium Networks this year with the recent announcement of Octeon III (Press Release) for the second half of 2012 and Nitrox III (Press Release), the new generation of Cavium’s crypto and compression accelerator.

Broadcom completed the acquisition of NetLogic Microsystems, Inc. (Press Release). It will be interesting to monitor how Broadcom will use NetLogic’s multicore processors in the future. Will Broadcom continue to sell NetLogic’s technology in merchant-market processors? Will it be used for the new generation of Broadcom networking chips that today already embed multicore technology? Both?

Platforms:

HP and Intel announced the SPECTRUM alliance. SPECTRUM enables developers of hardware and software to deliver solutions to meet the challenges of the telecommunications markets. (SPECTRUM Home Page). This alliance shows we are at the beginning of a revolution in the mobile and cloud infrastructure markets.

Mobile Infrastructures:

Alcatel-Lucent wins Best Infrastructure Technology Award for lightRadio Network.

Ericsson predicts 10x growth network traffic by 2016 (read here). Ericsson is currently ranked first in the LTE infrastructure market followed by Alcatel-Lucent and Nokia Siemens Networks (read here). You can also review the Samsung announcement claiming the company has set a goal to be a top three global LTE infrastructure supplier by the end of next year (read here). A real challenge knowing the Chinese big players (Huawei and ZTE) are also fighting to have a large slice of the cake.