IT Analyst Hype

Global data center traffic is set to grow fourfold, hitting 6.6 zettabytes a year by 2016. And cloud traffic, the fastest-growing component of data center traffic, will grow 44 percent annually, or sixfold, in the same time frame. So says Cisco's second annual Global Cloud Index.

"The biggest driver is that the network has become a more cost-effective storage repository. With computing in the old days, the cheapest way to store data was to put it on a local hard drive," said Zeus Kerravala, principal analyst at ZK Research.

New technology that helps direct the flow of information around data centers is becoming increasingly common in gear made by Cisco Systems Inc.'s competitors, challenging the Silicon Valley pioneer's dominance in its core router-and-switch market.

The concept, called software-defined networking, or SDN, allows companies to treat the mish-mash of equipment that underpins their data centers as if they were a single, seamless system. Many products use a standard known as OpenFlow, which analysts say could allow companies to buy more generic routers and switches and then reprogram them for their networking needs.

Using SDN could save companies money and enable them to employ fewer technicians. It's not known how big the market is for software-defined networking products because the concept attracts many different definitions, but major providers of business technology--including VMware Inc., Hewlett Packard Corp. and International Business Machines Corp. are moving aggressively into the area.

For its part, Cisco has approached the so-called SDN trend with a standard they say is more complete than OpenFlow. Executives there are taking a wait-and-see attitude toward adopting the OpenFlow technology, which could undermine the company's 68% enterprise-market share in routing and 74% hold on switches, according to research firm Gartner Inc.

Riverbed Technology Inc. announced new upgrades to Cascade, its suite of network performance management products that provide performance management visibility into virtualized infrastructures. It also introduced virtual editions of Cascade Profiler, Cascade Gateway and Cascade Shark for enterprises and managed service providers.

Riverbed Technology announced two new virtual editions of its Riverbed Cascade solutions, which enable managed service providers (MSP) to offer application-aware network performance management (NPM) as a service for small and medium-sized enterprise customers.

Juniper unveiled a new set of hardware and software products that run on the MX Series 3D Universal Edge router. By deploying the MX 3D, service providers can offer a number of revenue-generating, value-added services, such as load balancing, firewalls, content management, security services and content streaming right from the edge of the network. After a closer look, the range of services Juniper is offering is broader than what was available in the IP Service Switches of 10 years ago, but the concept is the same.

The MX 3D enables both consumer and business features including content streaming, security features and application network, as well as some operational tools in completely isolated, virtual instances as line cards within the router. So this raises a question: if so many companies failed at doing this before, why will Juniper be successful now?

I think the answer to this is that the industry is better aligned to network or cloud-based services. A decade ago, almost all enterprise IT infrastructure, and home infrastructure for that matter, was located on premise, so the concept of buying something like a firewall as a service was way outside the comfort level of corporate IT. Today that’s not the case. Cloud services are commonplace, and they are exactly what give network-based security and application delivery services a purpose. Also, when the concept of edge services was hot before, the value proposition was that a telco could tie these additional revenue-generating services to the other products it sells. What other products? An Internet pipe? That’s hardly game changing. Today if an enterprise is going to push a server into the cloud then it stands to reason that everything that surrounds the server should also be made available in the cloud. That’s the new telco opportunity. The shift to cloud computing opens the door for telcos to sell a wide range of other cloud services. The MX 3D is a much more cost-effective, scalable way of offering these servicescompared to deploying individual appliances on a one-off basis.

The one big challenge that I see here is the pricing model. Thinking back, the pricing model for something like a network-based firewall ranged from free to several hundred per month. Any work Juniper can do here to help its telco customers come up with pricing models will be well received, as I’m not convinced the service providers fully understand what to charge for these types of services.

F5 Networks, Inc., the global leader in Application Delivery Networking (ADN), today at VMworld® 2012 Europe announced support for VXLAN functionality, enabling organizations to seamlessly support software defined networking (SDN) initiatives by combining VMware and F5 solutions. F5® BIG-IP® products can natively take advantage of the benefits of VXLAN in the latest VMware environments, based on VMware vSphere® and vCloud Director®. In addition, F5 is announcing its intent to add VXLAN virtual tunneling endpoint capabilities to its BIG-IP products, providing the ability to natively encapsulate and decapsulate VXLAN packets. This innovation will further enable organizations to reliably scale cloud environments and streamline operations.

The Wi-Fi array, which I once described as either a super multi-radio AP or a collapsed-backbone WLAN switch (or both), is a great way to provision a lot of capacity in a given location, and Xirrus, who pioneered and lead in this space, have in recent years been offering arrays with modular radios. It's possible, then, to mix and match and even field-upgrade the radios in an array as new technologies appear. And that's exactly what Xirrus announced, a new pre-order upgrade plan for adding three-stream .11ac modules to their current product line. Xirrus arrays enable the collocation of 4, 8, or 16 radios in single package, and modularity is the key to investment protection as new technologies, like .11ac come along.

The only cautionary note here is that multiple .11ac radios could swamp the inherent backhaul capacity of certain models, so this will need to be taken into account when configuring a solution. Not all radios in a given array have to be .11ac, and it would be hard to recommend such anytime soon regardless, so Xirrus presumably has users covered.

I recently posted an entry on how I see the separate and distinct classes of wireless-LAN systems on the market today, with an open question as to the degree of functional overlap we'll likely see among them. The midmarket sector is today of particular interest, motivated by a recent announcement from Cisco of new APs (the 1600 and 2600) aimed just at that opportunity. Coupled with a new release of the Cisco Prime unified (wired and wireless) management console and an increasing emphasis on virtualizing key components (very important to midmarket firms trying to minimize costs), and the midmarket opportunity suddenly looks like anything but a niche.

We've identified the key requirements that define this segment. These include ease-of-use for operations staffs, because midmarket firms don't have the large IT departments of their enterprise counterparts. Staff productivity, always a key to lowering operational expense (OpEx), is thus especially critical here. The other major issue is scalability, as in non-disruptive, cost-effective, and easily managed, because midmarket firms often see higher growth rates than larger firms.

There's a good chance you've not heard of ConteXtream, and have spent little if any time thinking about software-defined networking (SDN). And, unless you're building an LTE network, you probably haven't thought much about the Evolved Packet Core (EPC), which essentially moves the heart of LTE networks into the IP world. But, trust me, what's happening here is huge - smack-yourself-in-the-forehead huge, in fact, applying key trends in networking overall to one of the most important, and most visible, areas of network investment today.

The company today announced an SDN approach to the EPC, in effect virtualizing the core of LTE networks. SDN, as you've probably surmised by now, is similar in concept to software-defined radio (SDR), the virtual WLAN controllers now available from a number of vendors, and virtualization in general - using commodity processors to replace specialized equipment. The cost reductions and flexibility inherent in this approach are, I believe, going to have a number of remarkable benefits going forward.

First of all, with costs going down, shortages of capacity will become rare. We won't have to wait for new developments from just a few vendors of such obscure but vital elements as SGWs and ePDGs. Reliability is improved via enhanced flexibility in load-balancing and meeting redundancy/failover requirements. Faster time to market - yes, that too, along with faster upgrades (even on demand; you'll likely be hearing much more about Layer-4 switches, for example). And perhaps even end-user organizations will be able to integrate with carrier networks, managing costs and provisioning services with remarkable ease. OK, I am likely a little bit ahead of reality with all of this. But I do see SDN as applied to carrier networks as radically changing the implementations of these vital capabilities, and much more, with perhaps even an IaaS approach dominating over time.

My big concern about LTE networks is that scarcity and expense are dominating carrier thought today - how else to explain the ridiculous pricing-by-the-bit plans now in place? I fully hope - and expect - that the carriers will meet their financial goals, with the magic of a competitive marketplace improving services and holding down costs for us mere-mortal end users. But, really, we need to move from a mindset of scarcity to one of abundance and possiblity - so much depends upon that, and I'm talking Layer-7 here. And I believe that SDN will make an important contribution to achieving that end, and may even be eventually recognized as the key development that made the ultimate success of LTE deployments around the globe possible.

Emulex Corporation announced the availability of its second generation of LightPulse® 16Gb Fibre Channel (16GFC) Host Bus Adapters (HBAs). The LPe16000B series of 16GFC HBAs are the only adapters on the market that offer PCI Express (PCIe) 3.0 support to complement the latest high performance Intel® Xeon® processor E5-2600 product family-based server platforms, industry-leading 1.2 million I/O operations per second (IOPS) on a single port, which is six times faster than current HBAs being offered by Emulex's nearest competitor, and BlockGuard™ data integrity offload for enhanced data protection.

IBM is making a bundle of product announcements that address the security challenges of mobile, cloud and Big Data computing, and I.B.M. is applying its own data analysis expertise to attack the data-security problem.

The I.B.M. offerings are mainly monitoring software for data flows to and from mobile devices, remote data centers and programs built with Hadoop, the underlying software for many Big Data applications.

I.B.M.’s moves reflect how the computer security business is evolving, said Jon Oltsik, an analyst for the Enterprise Strategy Group, a research firm. The industry, he said, has long been dominated by niche companies that focused on specific products, like virus software or firewalls. But the rising complexity of computing, and of security threats, is prompting a consolidation — and a need for products and services that span the spectrum of security threats companies face.

I.B.M., Mr. Oltsik said, is one of the few companies that has both the technology expertise and broad reach to be a winner. Other candidates, he said, include Hewlett-Packard, McAfee, Symantec and perhaps Cisco.

Enterprise Management Associates (EMA), a leading IT and data management research and consulting firm, today announced the release of its newest EMA Radar Report titled, EMA Radar for Enterprise Network Management Systems (ENMS): Q4 2012. Created to assist IT professionals in selecting the right ENMS products, EMA has identified the leading vendors in this space based on key criteria defined by EMA Managing Research Director, Jim Frey, and EMA Principal Research Analyst, Tracy Corbo.

The term “network management” encompasses a broad spectrum of solutions, ranging from single point products and element management to large enterprise-class solutions. For the purpose of this Radar Report, the term Enterprise Network Management Systems is defined as network-centric management solutions that are used by large organizations’ operations and engineering teams to discover, monitor, assess, troubleshoot, and generally maintain highly distributed enterprise networks. Evaluations were conducted based on extensive criteria in: deployment and administration, cost advantage, architecture and integration, functionality, and vendor strength.

“Managing the network and what is running on it has expanded beyond the domain of the NOC in the large enterprise,” said Corbo. “The emerging trend of ‘Everything IP’ is bringing more non-network devices onto the enterprise network. Server virtualization has added a whole new class of virtualized traffic, while BYOD and mobility has meant unplanned and unmonitored devices randomly jumping on and off the corporate wired and wireless networks. As a result, network monitoring and troubleshooting in today’s enterprise environments requires greater visibility and cooperation across IT domains.”

A detailed, comparative study of solutions from the following vendors is provided in the report: -CA Technologies -Centerity -EMC -Entuity -HP -Infosim -Ipswitch -Kratos Networks -OPNET -Paessler -Quest Software -ScienceLogic -SolarWinds -Zyrion

Results identify key strengths and weaknesses and highlight characteristics, summarized in a detailed market map and Radar Chart – which includes a composite score for each vendor – making it simple to see how vendors measure up in the market, as well as against other vendors.

Certain industries are already seeing an uptake in M2M, particularly where regulations are driving adoption.

Dan Bieler, principal analyst at Forrester, says there is a difference between the M2M and consumer environment and enterprise.

“From the consumer end it’s about having a lot of apps in the home, communicating with each other and elements of smart grids, such as electric and water meters. This includes applications in healthcare, monitoring heart rates which are linked up to a central database, so doctors can get in touch and inform patients they should have a check-up,” says Bieler.

“In the enterprise space there are a number of solutions and metering in the context of facility management. It’s being used to monitor areas such the oil industry, measuring holes in pipelines, asset tracking is another big area, as it is being used to track devices in hospitals.

“The range of applications is extremely widespread. It is quite horizontal in nature, being used to monitor, meter, navigate and notify in all sorts of sectors. But having said that, M2M surprisingly is still not a top priority. Our research found only one in 10 sees it as an important area.”

One reason for this is that the return on investment is often hard to demonstrate. In austere times the focus of the CIO is often on how to cut costs and reduce headcount: “It’s difficult therefore to come up with the required funding,” says Bieler.

“In some instances, companies are struggling to define which processes they want to support with that solution. But it is clear that interest is growing.”

One issue for the deployment of M2M is the lack of seamless connectivity for wireless and patchy 3G coverage. The GSMA says spectrum will be crucial in achieving a more networked economy, supported by a sufficiently flexible regulatory environment in the telecoms sector and in other industries. In the next four years the mobile industry will invest $793bn in expanding the coverage and capabilities of mobile networks, according to GSMA.

Networks provider Ericsson forecasts mobile data traffic will grow tenfold between 2011 and 2016. But even with new spectrum, mobile operators will need to be able to manage the fast-rising tide of traffic on their networks, both to deal with congestion and tailor delivery to specific service requirements, it warns.

Bieler says lots of small bits of data can add up to significant amounts, which can easily be underestimated. Smart metering in itself does not include a huge amount of data, but in theory it involves millions of meters sending information every 15 minutes.

Regular upgrades will have to be part of the broader M2M scenario, he says, raising the question of who should bear that cost. At the moment it looks like it would fall on carriers. “It will be a combination of developments, additional broadband technology, HTML5 will help and IPv6 of course. Plus the recognition that it can support and enhance business processes. There are a number of possibilities where M2M can play a role, such as automating processes that, in the past, would have been handled by people,” he says.

Bieler believes that M2M will eventually become embedded into business processes.

“I don’t see one part tipping the move to M2M, but several step changes, such as the 2014 EU regulations for cars that will certainly have a large impact on the automotive sector.”

Warren East, CEO of microprocessor company ARM – which manufactures embedded chips for M2M technology – recently told Computer Weekly the initial growth of this market will be slow. “Over the next five years I’m not expecting there to be much take-up – the opportunity will be over the next decade,” he says.

Unless you’ve been living under a rock, you know there’s no single technology trend that has more hype and mania around it than “software defined networks.” It’s this era’s “2.0,” or the technology equivalent of Andrew Luck. Lots of hype, plenty of potential, but has yet to produce anything. To date, almost every leading vendor has outlined an SDN strategy, but the one missing vendor has been the No. 2 market share leader, HP. After interviews with a number of HP customers and channel partners it appears HP is ready to unveil its SDN strategy.

From a vision perspective, there’s really nothing unique about what HP is doing - they’re focused on IT agility through the decoupling of control and data plane. HP has extended the value proposition of this to applications through a number of open APIs similar to what Avaya, Brocade and Cisco are doing. What’s not clear to me is who the application partners will be at time of launch, if any, but its support of applications is consistent with the rest of the industry.

Supporting the vision, HP announced OpenFlow support for the current 16 models of switches and released a new series of switches, HP 3800s that also have OpenFlow support. I believe there are 9 new stackables in the series, pushing HP’s number of OpenFlow-supported switches to 25. OpenFlow is widely supported today, so HP’s implementation of it is a logical move for the company.

To compliment the line of switches, HP plans to develop its own SDN controller, which appears to be scheduled for general availability in late 2013. This is a curious move as there are already so many controllers on the market. I’d rather see HP support as many controllers as possible, rather than build its own. However, as I’ve said before, the SDN wars have become a game of stacks, and HP is choosing to build an end-to-end stack. If you’ve got Cisco’s share this probably is the right approach, but HP’s No. 2 share position in networking is primarily made up of edge switching, meaning they’re a minority share player in the data center. HP’s best approach would be interoperability, which is something ProCurve has focused on since its inception.

The unique attribute HP can bring to SDNs, though, is a heavy services focus. HP has one of the largest services organizations in the industry and much of what will make a customer SDN deployment a success is the planning, deployment and ongoing management. Cisco’s approach is to train and certify its channel, whereas HP’s will be to leverage its massive service organization. This is the right approach for HP as its channel isn’t nearly as big or as well developed as what Cisco has. Cisco’s focus on channel partners, I believe, has greater long-term value, but HP could get some quick wins and rapid scale through its own services organization.

Cisco confirmed that it would halt production of its Application Control Engine (ACE), but the company says it may not be out of the load balancer business.

The changing nature of the virtual data center and the cloud have placed entirely new demands on load balancers and application optimization tools, and this leads Gartner Inc. distinguished analyst Joe Skorupa to believe that Cisco may be working on a larger data center solution in which a home-spun load balancer would only be one supporting element.

"We have already said that we believe Cisco will be in the storage market by 2015," Skorupa said. 

Ultimately Cisco could have an integrated storage, compute and networking solution and an application delivery controller could be part of that. 

Whatever Cisco decides to do, there is little time to waste before it loses any opportunity in the market. Within 24 hours of the news that Cisco would be ending ACE, load balancer rival F5 saw share value jump 8%, and A10 offered a Cisco ACE trade-in program that offered users who switched to A10 a rebate of $24,000 and a series of installation and migration services.

Cisco says its new system, the Nexus 3548, has latency measured at as little as 190 nanoseconds—billionths of a second—and is 60% faster than any competing products.