The skirmishes that have been taking place between Hewlett-Packard and Cisco as each tries to encroach on the other’s territory in an effort to own both the enterprise and consumer IT markets have been heating up over the past year. HP has been strengthening its ProCurve line of enterprise networking products, an area where Cisco dominates. Meanwhile Cisco has moved into one of HP’s traditional markets with the launch of its own enterprise servers, dubbed the Unified Computing System. That prompted HP to retaliate by inking a deal with Riverbed, long one of Cisco’s data networking archrivals.

As I’ve argued in the past, there are still other moves that HP could make to fend off Cisco in enterprise IT. But those would be rendered moot if Cisco decided to pull the ultimate offensive move — that of buying Dell.

The Math

Even in the face of the current economic downturn, Cisco could afford to buy Dell. It had some $41 billion in cash and assets as of the end of its latest fiscal quarter, though it would likely look to buy Dell using a combination of cash, debt and stock. The combined market capitalization of Cisco/Dell of around $150 billion would easily trump rival HP’s $96 billion.

The margins on Dell’s products are much thinner than Cisco’s have traditionally been (17 percent vs. Cisco’s 64 percent in their two most recent quarters, respectively), but combine the two and you have a very profitable business. My rough calculations — without taking any potential merger synergies into account — put a combined Cisco/Dell at $20.4 billion in top-line revenues per quarter with $7.3 billion in gross profits (36 percent). I am willing to bet that with synergies the combined company would have gross profits of over 40 percent — compared to HP’s most recent quarterly gross profits of just 24 percent.

The Products

In a move to further penetrate the consumer market, Cisco recently purchased Pure Digital Technologies, maker of the Flip Video camcorders. In the meantime, it’s seen good growth in its Linksys consumer networking products -– at least in areas where they compete with HP. However, while those products are a step in the right direction for Cisco, they won’t help the company win out against HP. But if Cisco bought Dell, it would get a wealth of consumer products that are directly competitive with HP’s, among them laptops, desktops, printers, digital cameras, monitors and more. Increasingly, all of these devices need to be networked together in the home, and Cisco has the technology to do that in spades.

Beyond the strategic fit on the consumer side that would result for Cisco, such an acquisition would make good strategic sense for Dell, too. Dell has always wanted to be a major player in enterprise IT, a traditional market for rival HP. To that end, the company has its PowerConnect enterprise networking products, and at the start of the year announced a partnership with Cisco focused on the next generation of data center switching — a partnership focused on battling the common enemy of HP.

In the meantime, Cisco unveiled its first enterprise servers with its UCS product line. Dell’s PowerEdge enterprise servers and blade servers would complement UCS and enable the combined company to sell a complete enterprise computing solution that could rival HP’s ProLiant and Integrity products.

The Result?

If Cisco wants to triumph over HP, the value proposition of a combined Cisco/Dell is compelling. Cisco/Dell would continue to dominate over HP in enterprise data and voice networking, would be a very formidable competitor against HP in enterprise servers and be well positioned against HP in the consumer markets.

The question is, what would HP do in response? Would it move to trump Cisco in the enterprise storage market by buying EMC and its crown jewel, VMware? Perhaps it would move up the enterprise IT stack and acquire Citrix for that company’s enterprise application suite? One way or another, if Cisco were to acquire Dell, HP would see it as a declaration of all-out war.

Microsoft is spending hundreds of millions of dollars to build out its next generation of data centers to host its cloud computing offering, Windows Azure Platform. While the company is clearly innovative in its data center designs and plans, the true reason behind its push toward the cloud may be its ability to turn a commodity product –- bandwidth –- into high gross profits. A quick analysis that we here at Panorama Capital did shows that the commodity business of selling the transfer of bytes may be one of the most profitable parts of running a cloud service.

Azure charges 10 cents for the bandwidth to upload and 15 cents to download a gigabyte of data. The disparity in pricing, I believe, is meant to encourage Microsoft developers to move lots of applications and associated data into Azure and then have lots of users access the applications from the same platform.

Let’s assume that a customer of Azure develops an application that downloads 10 gigabytes of traffic per day, 20 business days a month. That means the application downloads a total of 200 gigabytes of traffic in a month (and to make the point, let’s assume that the upload traffic is minimal). Azure charges the customer $30 per month for this bandwidth use (200 gigabytes times 15 cents per gigabyte), which seems like a small amount to pay.

If the customer’s application is only sending data and consuming bandwidth 12 hours a day –- all of its users are in North America — during 20 business days, that means that the customer is effectively using 1.85 megabits per second of bandwidth during the month (200 gigabytes per month divided by the sum of 20 days times 12 hours in a day times 60 minutes an hour times 60 seconds per minute). Put another way, the customer’s $30 per month equates to a bandwidth charge of $16.20 per megabit ($30 divided by 1.85 megabits per second).

Cloud service providers buy a lot of bandwidth to provide access to the Internet. My market research (albeit not exhaustive) puts the current price per megabit of bandwidth for a large cloud provider at around $8. That means that for the 1.85 megabits per second that the customer uses, Azure effectively pays $14.81 to its bandwidth provider and keeps $15.91, or 51 percent gross profit — not a bad profit margin on a commodity business like bandwidth. To be fair, other cloud service providers, like Amazon and Rackspace, charge similar or higher bandwidth fees and likely make similar gross profits.

While this example is for a fairly limited charge to the customer of Azure, if a customer were to build an application on Azure that generates a lot of bandwidth, then the costs and profits to Microsoft get substantial. For an application that downloads 200 gigabytes per month, the costs are $600 for bandwidth for about 37 megabits per second of usage (again assuming the application only consumes bandwidth half of the day for 20 business days a month). Of that $600, Microsoft pays its service provider $296.30 and keeps $303.70 of gross profit.

And the model scales linearly –- if Microsoft can build up Azure to cumulatively charge for all customers’ applications sending 100,000 gigabytes of data a day, it will reap approximately $151,000 in daily gross profit off bandwidth. That would equate to a lot of copies of Microsoft Windows sold daily without any packaged software developers on staff required. With these gross profit margins, one can easily understand why Microsoft is building thousands of megabits of bandwidth to serve data from Azure applications.

So if you’re using a cloud customer and your application will send a lot of data from the cloud, our analysis indicates that once you’re sending over 50 gigabytes of data daily (or a terabyte a month costing you $150 on Azure, for example), it may make sense to leave the cloud and buy your own bandwidth to the Internet –- you’ll probably save 50 percent of your monthly bandwidth charges. The trick will be moving your application from the cloud to your own infrastructure and dedicated bandwidth and then finding the expertise to manage this environment. Cloud service providers are counting on that being a difficult trick to perform.

When Cisco Systems announced plans to enter the enterprise server market, no company stood up and took notice more than Hewlett-Packard — the HP ProLiant line of servers, after all, is a force in the enterprise market; Cisco’s entrance was essentially a declaration of war on its former partner. To paraphrase one of my favorite characters on “Lost,” “They changed the rules.” Now, HP must change the rules again, for in order to win this war they are going to have to first win numerous strategic battles, and for that they’ll need to start shopping.

HP needs to first realize that the focus of this war is the enterprise IT market. As far as the other markets Cisco serves, service providers are already being flank attacked by Huawei, Juniper, Alcatel-Lucent and Ericsson, to name just a few, while DLink, Netgear and dozens of others are busy in the Linksys consumer markets.

HP needs to think about the markets their enemy is headed toward, not where it has been or is today. They need to think about what the IT marketplace will look like in 2-3 years, when CIOs could actually have budgets again. Startups are where the innovating ideas of tomorrow are already being developed, and for many of these companies, their price tags may never be this low again.

So, which startups should HP acquire? If it’s looking to make gains at the expense of Cisco, the list is as follows:

Cavium Networks: This Mountain View, Calif.-based startup builds intelligent processors for networking functions such as compression, encryption and data acceleration. Notably, however, Cisco is a major customer of Cavium’s, so I imagine they’re contractually obligated to provide Cisco processors for many years to come.

Arista Networks, Blade Network Technologies or Force10/Turin Networks: While HP’s ProCurve line of switches has become a significant player in the enterprise data networking market, it needs to have a winning 10/100 gigabit Ethernet switching product line. Each of these startups have Ethernet switching technologies that could help HP aggressively attack Cisco at the core of the data center network.

DataDomain, StorWize or Ocarina Networks: Storage optimization is the emerging technology that CIOs need in order to cost-effectively deal with the mountains of data that enterprises store over long periods of time. With an $800 million market capitalization, DataDomain may be too large of an acquisition right now, but both StorWize and Ocarina Networks are just the right size.

BlueCoat, Zeus Technology and Strangeloop Networks: The application delivery market, focused on making applications run faster and optimally, is a battlefield strewn with many players. While HP recently inked a partnership with Riverbed, this is merely a shot across the bow. HP needs to acquire in this area, and although market leaders F5 (market capitalization of $2 billion) and Riverbed (market capitalization of $800 million) may be too expensive, BlueCoat, which has a $400 million market cap, may give them the firepower they need. Likewise, startups focused on the next generation of application delivery such as Zeus Technology and Strangeloop Networks would provide the necessary weaponry.

Nirvanix or Parascale; Rightscale or Elastra; SkyTap: Most of the companies in the enterprise cloud computing space are still privately held, so this may be the ideal time to make decisive acquisitions. HP should acquire companies focused on helping enterprises reach storage and compute power in the cloud in a scalable manner, those enabling what I call the Cloud Two-Step. In the storage battleground, HP should move to provide outsourced storage clouds, perhaps by acquiring Nirvanix, or allow enterprises to build internal storage clouds by buying Parascale. In the compute cloud market, HP needs to acquire a company such as Rightscale or Elastra. To help enterprises build and scale applications, SkyTap, which provides testing resources in the cloud, would make a compelling acquisition.

To win the upcoming war, HP needs to pick its battles carefully and strike quickly. For the past decade, Cisco and HP have had a cold war alliance –- Cisco claimed enterprise IT networking while HP focused on enterprise servers and compute. Both companies have been rewarded with exceptional market capitalizations. Now that Cisco has broken that alliance and declared war with their enterprise servers, HP must respond. I suspect bloodshed will soon ensue, so HP should move quickly — or risk losing the war before the first shot is fired.

It seems as though our Canadian neighbors have imported our “too big to let fail” economic policy. The Canadian government on Wednesday pledged to give telecom equipment maker Nortel Networks, which has suspended efforts to sell its Metro Ethernet unit, up to C$30 million ($24 million) to emerge from bankruptcy. The decision to use taxpayer money to save the telecommunications equipment giant has generated a lot of heated discussion, in the private equity and IT communities both in Canada and stateside. But is it the right one?

Nortel has long been considered the technical jewel of the Great White North. The company’s origins date all the way back to 1895; it was responsible for inventing and deploying many fundamental telecom technologies, perhaps the most prominent example being the digitizing of phone calls. Its most successful product, the DMS-100, was a digital central office switch that provided telephone service for some 100,000 phone lines. The name Nortel has evolved as well: The firm began life as Northern Electric, then went on to be known variously as Bell-Northern Research, Northern Telecom Limited and finally, after its merger with Bay Networks, Nortel Networks.

But is preserving the country’s technological heritage reason enough to spend millions in taxpayers dollars? I am not an economist, but rather a venture capitalist, and from that standpoint I could argue that the telecommunications market needs strong global competition to be successful and that the absence of Nortel would leave a notable hole in the marketplace. On the other hand, Nortel has clearly suffered from poor leadership and operational execution, so perhaps the market has spoken — and we should just listen.

All of which, of course, sounds eerily similar to the arguments here in the U.S. over the government-led bailouts of certain financial and auto companies. On that note, one small piece of advice for Nortel executives as they work with the Canadian government on their bailout: Don’t fly the corporate jet from Toronto to Ottawa.

Microsoft Data Center Chief Mike Manos posted a blog entry yesterday on the company’s vision for next generation data centers. The blog post (and the accompanying animated video) has extensive details on how Microsoft envisions building the data center of the future — and it definitely has some of the “trailer park” modularity and scalability attributes that I mentioned in my post last week.

Some of the key features of the Generation 4 Modular Data Center design are the use of an innovative spine infrastructure for cooling, power and connectivity; the use of pre-fabricated and completely modular server containers and buildings; and ambitious goals for energy efficiency (an average power usage effectiveness (PUE) of 1.125 for all data centers by 2012).

By putting forth its plans for the Generation 4 Modular Data Center, Microsoft is simultaneously throwing down the gauntlet in the data center arena and showing its cards. By openly sharing its data center plans it is taking the technology and thought leadership away from Google in the critical area of Internet and services infrastructure. At the same time, in somewhat typical Microsoft fashion, it is asking (or perhaps requiring) the industry to standardize on its data center design approach. Given its influence and buying power with vendors it may get its wish.

Viewing this information in conjunction with announcements of the Azure Services Platform and Office Live, there is no doubt that the giant in Redmond is aggressively focused on delivering enterprise cloud computing. To top it off, market timing could not be better for Microsoft, as its hordes of cash and diversity of enterprise products should prove one version of the Golden Rule: “He who has the gold makes the rule.”

Related stories:

• Inside Microsoft’s Internet Infrastructure & Its Plans For The Future
• GigaOM Interview: Ray Ozzie

]]> http://gigaom.com/2008/12/03/microsoft-reveals-fourth-gen-datacenter-design/feed/ 4 gigaallanleinwand image-thumb http://gigaom.com/2008/11/26/trailer-park-20-where-all-your-data-lives/ http://gigaom.com/2008/11/26/trailer-park-20-where-all-your-data-lives/#comments Thu, 27 Nov 2008 00:25:20 +0000 Allan Leinwand http://gigaom.com/?p=30382 ]]>

The prospect of outsourcing servers and storage to the cloud has an irresistible lure of operational simplicity and cash efficiency for today’s application developers. Cloud computing vendors help operate social networking applications, micro-blogging sites, global gaming networks and a plethora of applications that we use everyday. Yet, as successful and economically desirable as clouds have been for many organizations, outsourcing servers and storage causes a serious emotional and operational dilemma for the hardened breed of systems administrators called server huggers.Everyone working in and around the Internet knows a server hugger. Server huggers relish spending time in air-conditioned data centers, sitting on raised floors under florescent lighting with a laptop connected to a console port of a server (or, if they are lucky, standing against a server rack using a dedicated terminal and a slide-out keyboard tray). They spend hours staring at command-line on a terminal and at notebooks of commands, passwords and IP addresses.

Server huggers like being near their servers -– they feel that the emotional well-being and efficient operation of their servers requires them to be physically close at all times. And while it may be easy to scoff at server huggers and their technical idiosyncrasies, it is this same breed of folks that keep the applications running on the Internet — we would all be grasping to master our social graph, unable to micro-blog to our abundant followers and forced to live in our real life without them.

But server huggers face an impending crisis — the data centers that host their servers in many large metropolitan areas such as Chicago, Los Angeles, New York and San Francisco are increasingly filled. It is unfulfilling to hug a server that sits in a cardboard box because there is no rack space left in a data center, so server huggers have been scrambling to put their servers in geographically desirable locations that do not require new construction or an exorbitant budget.

And that got me thinking: Does the data center of the future look like a mobile home park? A mobile home park provides a place for you to park a single-wide or double-wide home and some basic utilities — power, roads, mail, etc. Yet, unlike seemingly every person on the Jerry Springer Show, servers do not operate well in mobile homes. However, as Microsoft, HP, Verari and others have shown, high density blade servers can be packed with hundreds of terabytes of storage, cooled and operated efficiently inside standard shipping containers. Maybe instead of more metropolitan data centers for the server huggers, we need container parks.

A container park would be on a plot of land within a metropolitan area (or on the immediate outskirts) and provide basic services to host containers filled with servers. Container parks could be located in spare lots close to power generation facilities, or be set up along a high-bandwidth fiber routes or even adjacent to a telecommunications facilities.

Compared to building, a top-tier data center that can cost $1,000 per square foot, setting up a container park could be done relatively cheaply — all is needed is a plot of land with the appropriate physical security, a power distribution plant, backup generators and abundant Internet connectivity. These items are available in metropolitan locations where server huggers and their employers congregate. While the containers themselves are self-contained, there should be no reason that multiple organizations and their server huggers could not share servers in a single container. After all, server huggers already share cabinets and cage space in data centers today.

So, will container parks soon emerge as the next bastion of the server hugger? Given the choice of moving their beloved servers to the cloud or hugging them in a nearby cool container down the street, the choice seems somewhat obvious.

Image courtesy of Verari.

Anyone who knows me will tell you that I am not a great dancer. Yet I have recently become a proponent of a technology strategy that I have coined the “cloud two-step.” The cloud two-step is the way enterprise IT departments are going to dance their way to better infrastructure economics for their organizations during the current downturn.

Before we talk about specific two-step moves, it’s important to realize that the economic crisis has forced enterprise IT departments to stop or severely limit their spending. While this may help the enterprise stay afloat, it also requires that those departments find imaginative and cost-effective ways to serve technology needs. Put another way, they need to find the cheapest way to provide two commodity resources: storage and processor power. And providing flexible, cost-effective and outsourced storage and processor power is the value proposition of cloud computing that makes the CIOs of Fortune 1000 companies stand up and take notice.

Enterprises are looking to leverage the buying power and infrastructure management expertise of cloud vendors such as Amazon, Microsoft, Google, GoGrid, Mosso, Joyent and others. These clouds allow the enterprise to lease rather than own processors and storage. The lure of leasing these resources allows for the flexibility to turn up and down resources as needed instead of owning and provisioning for a high watermark of requirements. Perhaps one of my favorite illustration of the power and economics of cloud computing is the Washington Post example available on Amazon.com.

So why are enterprise IT departments not running toward cloud computing to help scale resources and control costs? One oft-cited reason is that this technology is still in its infancy, but I think the real reason is that they are struggling to transform their traditional practices to ones that extend into the cloud. For example, if an IT department decides to outsource some of its processing power to Amazon like the Washington Post did, then how are those resources provisioned, monitored for faults and performance issues, secured against break-ins and accessed by both internal resources and third-party partners? There are some cloud solutions for these issues, but enterprise IT has spent decades building integrated internal systems and needs a way to bridge the gap from what they are using today to the cloud.

That’s where the cloud two-step comes in. The cloud two-step is a technology strategy that bridges the gap between traditional IT practices, ones that occur within the four walls of the enterprise, and the desire for IT practices that work under a new economic environment. It is a way to step into the cloud in two easy steps: the first to deploy technology within the enterprise that IT can fully control, the second to then use this technology to leverage clouds. Some examples include:

• a technology solution that configures enterprise network management systems (step one) to monitor faults and performance of cloud vendors (step two)
• software that configures virtual machines in an enterprise data center (step one) with an overflow option to offload processing to a cloud (step two)
• a new enterprise backup system (step one) that automatically provisions cloud storage when disk capacity reaches a certain threshold (step two)

It seems likely that there will be many dancers auditioning to perform the cloud two-step for enterprise IT in the upcoming year, and I look forward to seeing these solutions help further establish clouds as the next generation for IT. So while enterprises will always need their IT departments, those that find solutions that allow them to dance the cloud two-step gracefully will live to dance another day. Those that don’t may find that their next dance is their last one.

Allan Leinwand is a venture partner with Panorama Capital and founder of Vyatta. He was also the CTO of Digital Island.

There has been a rumor rustling around Silicon Valley for a number of months that Cisco Systems is on the verge of acquiring EMC. Such a move would make a lot of strategic sense for Cisco, but this rumored mega-merger of technology giants may have to wait for the U.S. economy to recover before becoming a reality.

If Cisco were to acquire EMC, it would have an enormous impact on the technology landscape and etch in granite the combined company’s role as the hub around which the rest of the enterprise data center industry revolves. It would also place the firm at the forefront of the ongoing synergy between storage and data networking, a trend observed back in July, when Brocade Communications Systems agreed to buy Foundry Networks for $3 billion dollars.

And it would give Cisco control over VMware, the leader in enterprise virtualization software, and help move it further up the technology stack from being a data networking vendor and into enterprise software. Last month at VMworld, Cisco announced the Nexus 1000V switch, an integration of their switching software with VMware ESX (this was not what I predicted, but it was a step  toward my prediction).

Such a deal could also help Cisco satisfy Wall Street by growing revenues, something that the company has done very successfully through other acquisitions, although not on this scale, in the past. This is where the economic crisis comes into play. Cisco had $26.2 billion in cash and cash equivalents as of its most recent financial quarter. EMC has a market capitalization of around $24 billion. Taking into account its current cash balances and outstanding debt, EMC’s enterprise value is around $22.4 billion. If Cisco wanted to pay a premium to EMC shareholders to purchase the company it would more than likely need to pay at least $25 billion or slightly less than its current total cash on hand. While Cisco is generating billions of free cash every quarter, spending all of its cash to acquire EMC would not be a prudent fiscal move.

If this rumored acquisition was considered strategic enough to the Cisco board of directors, one way to fiscally accomplish it without draining the cash reserves would be to raise debt. But given the current economic crisis, the debt vehicles that are available on the market are either non-existent or on very unattractive terms. So for now, it seems likely that Cisco will need to put its rumored acquisition on hold until the economic crisis gets resolved.

Allan Leinwand is a venture partner with Panorama Capital and founder of Vyatta. He was also the CTO of Digital Island.

Recently two separate posts ran here that argued in favor of technology entrepreneurs who need early-stage money taking it from angels. I, however, believe they should seriously consider taking it from people in my profession: venture capitalists. The current economic meltdown and its effects, which are expected to be felt for at least the next year, require investors with deep pockets and a long-term investment horizon starting from a company’s earliest days.

Entrepreneurs seeking funding usually turn to one of two sources: angels and venture capitalists. And given that I fall into the latter category, let me go on record as saying that both sources can provide exceptional outcomes. In fact, in many cases, some of the best deals that we see in venture capital come from angel investors. So angels and venture capitalists are not mutually exclusive sources of funding, as one can lead to the other. However, with the economy as it stands today, entrepreneurs need to think clearly about how to solve their funding needs from now all the way until better economic times.

I have known many entrepreneurs that have worked with angel investors and had great outcomes. Generally, angels are exceptionally smart entrepreneurial individuals who make fast investment decisions, take a smaller ownership percentage than venture capitalists in startups they fund and opt to have minimal involvement in the operational aspects of the business. On the other hand, some entrepreneurs note that angels are not focused enough on their business (anecdotally I have heard of some angels spreading their investments over 30 or more startups) and some have a fairly short expectation on the investment horizon for returning their capital.

For an entrepreneur, working with a venture capitalist is a different experience. Investing our limited partner’s money generally requires us to have a more diligent and lengthy investment process than angels, to work with a very limited set of companies (at Panorama Capital we limit our full-time involvement to 5-7 companies for each partner) and thoroughly study a market before investment. As venture capitalists, we would argue that this focus and investment commitment should equate to more ownership in a startup than is typically taken by an angel investor.

After the initial investment, we also give access to syndicates for additional financing, study the market landscape for partners and competitors and help with team building. In other words, while we are more focused on our investments, we are also working on behalf of our limited partners to provide a return on their investment and that, in some instances, can admittedly result in a conflict of interests between us and the entrepreneur. In my experience, this conflict happens less often than one might think, typically when there isn’t a strong and productive relationship between the investor (angel or venture capitalist) and the entrepreneur.

While there are objective reasons to take capital from both an angel or venture capitalist, in my mind the current economic downturn means that an entrepreneur needs an investor who will commit to a startup over a much longer time period than a typical angel investor. For example, before we commit, we ask ourselves if this is an investment that we would be excited about being involved with for 10 years.

The technology IPO market is dead and many predict it will stay that way for at least another 18 months. Debt financing for larger firms to use for mergers and acquisitions is practically non-existent. So unless you’re an entrepreneur working at a startup that is narrowly focused on getting bought by a company that is flush with cash (and there are a few of these, such as Google, Microsoft and Cisco), you’re going to need an investor with a long-term commitment to your business.

While I do know some truly remarkable angels who can help on all of these fronts, overall it seems clear to me that now, more than ever, is the time for venture capital.

Allan Leinwand is a venture partner with Panorama Capital and founder of Vyatta. He was also the CTO of Digital Island.

Cisco Systems will support VMware virtual machines on their networking hardware? There’s buzz around Silicon Valley that there will be a big announcement made at VMworld next week in Las Vegas, and that’s my prediction as to what it will be. The integration of virtual machines and networking, which was signaled last year when Cisco invested heavily in VMware just prior to the virtualization company’s IPO, would have numerous ramifications, not only for the two companies, but the networking industry overall.

If my prediction comes true, it would help Cisco remain relevant in the data center, allowing it to do more than move IP packets between servers. It would also entrench the company into the enterprise, distancing themselves even further from the likes of Juniper Networks and 3Com, both of whom have struggled against Cisco to gain some toehold in the enterprise infrastructure marketplace.

And this move would allow Cisco to participate in the server and virtualization ecosystem itself. Organizations that have invested heavily in Cisco infrastructure would be able to expand the lifetime of their equipment by running on their networking equipment any software application that runs in the VMware virtual machine — a set of applications that’s growing every day. Enterprises that want a single vendor to supply both their networking infrastructure and their virtual machine environments, meanwhile, could reap benefits from integrated support. As a benefit to Cisco’s internal operations, this integration would allow them to move into higher margin application support, possibly allowing for easier acquisition integration.

The downside for enterprises that use VMware virtual machines on Cisco hardware is that they’d be unable to leverage the Intel x86 server hardware ecosystem to their advantage. Today, without the VMware and Cisco integration, enterprises buy hardware to run their software applications that have benefited from the large economies of scale in the Intel supply chain. Combined with the effect of Moore’s Law, this has resulted in an excellent price-to-performance ratio for enterprise compute platforms. These performance increases in the server hardware rise in tandem with market demand and volume and are not bound to the high-margin expectations of products like Cisco’s.

So when tied to Cisco hardware, which would have lower volumes than the Intel x86 ecosystem, this solution would inevitably result in higher price and less performance. Cisco has always been a hardware company and this move could tie enterprises into their higher-margin products. Indeed, in some instances, Cisco hardware that performs commodity functionality has been sold at over 40 times the price of commodity hardware that performs the exact same function.

While this is not the Cisco Blade Server that I forecast earlier this year, if this prediction does come true, it will undoubtedly cause additional friction between Cisco and the giants of the server industry, namely IBM, HP and Dell. If enterprises can run applications in VMware on Cisco hardware, it could obviate the need for some servers to exist.

Finally, with this step networking would evolve from being the technology that moves bits between servers to participating in the central role at the heart of the enterprise data center. Competitors in both markets — server hardware and networking — should take note and prepare for further waves of change in the enterprise data center.

Allan Leinwand is a venture partner with Panorama Capital and founder of Vyatta. He was also the CTO of Digital Island.

Broadband service providers are looking to add higher-value services to their offerings, services that could soon include a virtual desktop for consumers. Indeed, the idea of a service provider offering a PC as a Service (PCaaS), essentially a PC in the cloud, may be coming to your broadband connection sooner than you might think. Here is how a virtual desktop would work: You’d have an access device at your location, called a thin client, which would connect your keyboard, video screen and mouse (KVM) to the service provider’s broadband network. (For more detail on thin clients, see Stacey’s recent post.)

The thin client could be a hardware device or it could be a piece of software running on your current PC. In either implementation, the thin client sends all of your KVM data from your location to a server hosted in the service provider’s network. All PC functions and applications would be running on the server in the network and the only data going between your location and the server would be KVM information.

That is the major benefit of a virtual desktop: All operating system files, applications, documents, security software and so on are located on the server. All you need at your location is the thin client and you get access to your full desktop. These benefits, however, also highlight the main drawback of a virtual desktop: lack of portability. Moving your data from one virtual desktop to another may not be a trivial task and some applications may not be portable into a virtual desktop at all.

The technology to offer a virtual desktop has been around for a number of years. Companies such as Citrix, VMware, Microsoft and others already provide software to virtualize user desktops and connect to thin clients. The main issue with these offerings has been their performance relative to local computers. I, for example, was subjected to the horrors of using a software-based thin client connected to a server in a remote location, and the performance was abysmal at best.

But two fundamental technologies that may solve the virtual desktop performance issues already exist. The first is the proliferation of broadband Internet. Using a thin client to connect to a remote virtual desktop server over a multimegabit link that is within the same metropolitan area can provide reasonable performance. The second technology is KVM enhanced by hardware. Companies such as Teradici and Pano Logic provide hardware acceleration and compression for the KVM data passing from a thin client to the server — to the point where the performance difference between a local computer desktop and a virtual desktop is nearly indistinguishable. Using these technologies, the performance of the virtual desktop could even provide a graphics-intensive experience, such as playing a 1080p HD movie in one window and playing an action-packed game in another.

Assuming that the technologies exist to enable service providers to offer virtual desktops for consumers, from a business perspective, PCaaS has numerous appealing qualities. Broadband customers that use a virtual desktop will more than likely pay for a higher-bandwidth broadband service. Given the portability issues around virtual desktops, this also provides a clever mechanism to lock the consumer onto a specific network, which would ostensibly result in lower churn. I can also envision different product bundles for consumer-focused virtual desktops: a basic desktop with a browser only, an enterprise desktop with Microsoft Office applications enabled and a gaming/HD desktop that comes bundled with a hardware-based thin client branded by the service provider (Here’s your AT&T desktop access device!). Additional options could be storage space, accessibility options (Do you want to access your desktop from any TV and your mobile phone? Please pay us $5 more per month) or peripheral device support (such as printers and webcams).

One question that needs to be answered is how service providers could offer virtual desktops in conjunction with their metered bandwidth services. If I am sending lots of KVM data to a virtual desktop hosted by my service provider, you can be darned sure I don’t want be billed on a bandwidth meter.

But perhaps the biggest unknown around the PCaaS business is the user support that would be required. Service provider support organizations are better known for frustrating their users than helping them. Extending these support organizations to answer a myriad of desktop, application and device peripheral issues might be too much for them. An alternative may be for service providers to outsource the application and peripheral support to someone like Microsoft, similar to today’s relationship between MSN and Qwest.

With all of this to consider, are you ready to give up the hassles of managing your own desktop for a virtual desktop run by your service provider?

In 10 years, which company will own the cloud computing space? That question has been the subject of long and contentious funding debates over the past year, especially here on Sand Hill Road. And while I know that predicting the future is an inaccurate science, I think that when it comes to evaluating this nascent industry, the VC community has been focusing on the wrong criteria.

Today, cloud computing offerings are application-specific frameworks that are run by companies both large and small. Google’s App Engine is a cloud for running Python applications; EngineYard is a cloud for Ruby-on-Rails; Amazon’s EC2 and S3 provide generic compute and storage clouds, and so forth. While each of these companies addresses a vertical market need, I believe that by 2018, clouds will instead be evaluated based on three generic criteria: transactions, user experience and presence. And as with any active market, it’s a safe bet that there will be plenty of companies that best showcase each of them. One of the emerging trends in cloud computing is providing infrastructure that allows businesses to perform transactions. Two companies that immediately come to mind are eBay and Amazon; both have significant infrastructure expertise and business divisions devoted to processing their customers’ transactions. Amazon in particular just launched Checkout, which facilitates highly scalable transaction services. Imagine if over the next 10 years these companies find a way to bring their scale and transaction expertise to a cloud offering across multiple industries and market segments.

What will become increasingly critical is providing cloud consumers with a spectacular user experience, something nobody does better than Apple. While I realize the company is not strictly focused on cloud computing today, imagine a scenario in which it leverages the success and elegance of iTunes — essentially a cloud for selling digital media — to other markets. I can foresee a cloud computing environment where Apple allows users to build applications using their user interface templates and designs; every application developer that strives to make their application “as sexy as iTunes” could leverage this infrastructure. A current example of Apple’s approach can be found in its new MobileMe cloud, which emphasizes ease-of-use and user experience for keeping email, calendar and contacts synchronized. Elsewhere, Google Gears and Microsoft Live Mesh are also attempting to be environments in which users develop cloud applications, but they don’t seem to have the same focus on consistency of user experience.

Speaking of Microsoft, when it comes to presence in the computing space, they have an enviable position. With their software presence on the PC, mobile phone, game console, media center and even autos, they’re set up to be everywhere for at least the next decade. And the Redmond giant has recently changed its PR to emphasize software and services, which leads me to think that they’re moving to offer a wealth of cloud-based services.

The key will be leveraging their almost ubiquitous presence across nearly every aspect of the computing space to convert their hordes of desktop and IT application developers to work on their cloud in order to come up with future services for the same markets. If Microsoft can execute on this strategy, use their presence and motivate their developers, they will be a significant player in the cloud computing market in 2018.

It seems clear that the dominant cloud computing company in 2018 will be able to process transactions on the scale of Amazon and eBay, have the eye-popping user experience of Apple and the presence of Microsoft. Which cloud company do you think it will be?

Fellow GigaOM writer Alistair Croll contributed heavily to this post.

Here’s a horror scenario for everyone on the content side of the Internet: A consumer comes to a web site to download a movie, work presentation, software update or photos, and just before they commit to the download they pause and wonder: Am I over my usage quota this month? How much will downloading this new HD movie from Netflix on my Xbox cost me?

We’ve all been there before — with cell phones, about a decade ago. Usage-based pricing tiers started out with very limited minutes and lots of overage charges. Competition in the market by innovative operators drove plans fairly quickly to a point where only exorbitant usage resulted in overage charges (and now there are flat-rate plans for those consumers, too).

Unfortunately, the usage-based pricing plans (starting at 5 gigabytes) being considered by AT&T, Time Warner and others will force us all to wonder about the size of our connectivity bill on a monthly basis. Further, the lack of last-mile (the infrastructure that connects the consumer to their Internet service provider) competition will not result in these plans changing in the near future. Today, true competition on the Internet last mile requires new copper or fiber to each consumer — a very costly proposition. Cellular competition, on the other hand, required a less costly (on a relative scale) deployment of cellular towers.

While it is true that the consumer can elect who provides services over their last mile, most of us have very limited choices. As an example, a friend of mine recently moved into a building in downtown San Francisco that had exactly one last-mile provider: AT&T. The 700Mhz wireless spectrum provided a hope for an alternative consumer last-mile option, but that dream quickly faded.

Competition and an aggressive last-mile build have resulted in reasonable usage-based pricing models in Japan. OCN, the carrier operated by NTT Communications, is planning for unlimited download bandwidth usage and a 30-gigabyte limit on daily upload usage capacity. By my estimates, that will be more than adequate for all but the largest consumers of Internet bandwidth and does not invoke any horror scenarios for the large content owners.

In fact, large content owners may help us all avoid usage-based pricing horror scenarios. They spend hundreds of thousands of dollars every month (assume $10/month/Mbps using 95th percentile on 10Gbps of traffic) with the same Internet service providers buying connectivity to their networks because they want to be connected directly to the consumers via the last mile.

If the Internet service providers start billing on usage-based pricing, it’s inevitable that large content owners will look for new ways to reach the consumer. It seems unlikely that they’ll be willing to pay the service provider for access to their last mile if at the same time the consumer is being motivated not to access their content. Why would Microsoft and Netflix pay Time Warner for connectivity to their cable Internet infrastructure consumers if those same consumers are being billed on usage and worry about their usage quotas before downloading HD movies onto their Xbox?

Like other large businesses, Internet service providers are looking for ways to extract more value from their customers. As a venture capitalist, I understand and appreciate that perspective. Usage-based pricing, however, at least as currently envisioned by the service providers, will not only change consumer behavior but will work against some of their larger customers.

According to AT&T, Time Warner and others, usage-based pricing is coming to your Internet connection. While the reasons for this change in pricing model are varied, both in terms of technology and politics, it’s clear that consumers used to an “all-you-can-eat” buffet of streaming video, photo-sharing and podcasts are headed for a lean diet of Web 1.0 and email. Unless, of course, you want to pay a lot more for your Internet connectivity.

How much more? While the service providers have not announced their pricing plans, it seems clear that usage-based pricing will be based on the number of bytes you send and/or receive from the Internet on a monthly basis. Time Warner has suggested that usage-based pricing will be tier-based, with tiers at 5, 10, 20 and 40 gigabytes and overage charges applied for bytes that exceed them.

To put those numbers in perspective, here in the Bay Area I subscribe to AT&T DSL for $24.99 per month. I can download at 1.5 megabits per second and upload at 512 kilobits per second, which means I am bit-rate limited to downloading 500.2 gigabytes per month, or about 20 gigabytes per dollar. That same $24.99 per month also allows me to upload 165.9 gigabytes per month, or about 6.6 gigabytes per dollar. But to keep the pricing simple, let’s assume that I’m currently paying 5 cents per gigabyte sent or received. Granted, I may not consume all of these gigabytes every month, but in theory, I could.

I think it’s safe to assume that the service providers will price their usage-based tiers at amounts comparable to today’s monthly fees. They’ll want to lure in customers to the lowest price tier and then gouge them with overage fees. So let’s assume that the lowest priced usage-based tier, 5 gigabytes, costs $10 per month. That equates to an increase in my current fee of 40 times, to $2 per gigabyte. The highest tier, 40 gigabytes, will undoubtedly cost the same or more per gigabyte. If we assume that this tier will be priced at the same cost per gigabyte, then that equates to $80 per month. And again, that’s without overage fees, which will undoubtedly be as hefty as the surcharges on cell-phone plans.

As a rough reference, 5 gigabytes is the equivalent of doing one of these activities over the course of a month:

• downloading about 1,000 songs from iTunes (assuming about 5 megabytes per song)
• downloading five full-length movies from iTunes (assuming a two-hour movie)
• watching about 500 minutes of YouTube video (a quick test I just ran shows that a 2.5-minute video is a 5-megabyte download)
• sharing about 2,500 two-megabyte pictures (as normally produced by today’s typical 8-megapixel camera)

These references are estimates and do not account for other ways we typically use bandwidth during a month, among them file backup and recovery; VPN connections to the office; IP video conferencing; downloading Microsoft software upgrades and patches; use of cloud computing sites such as Google Docs and Amazon’s EC2; and so forth.

Of course, service providers will argue that in reality I do not consume 500.2 gigabytes of data each month, that my effective cost per gigabyte is higher than 5 cents and closer to the usage-based prices. And if I’m only browsing the web, doing email with small attachments and downloading the occasional picture, then my usage should fit in the 5-gigabyte usage tier and my monthly bill could actually go down. But that’s not the point — the point is that the unit economics of the Internet have changed and consumers are going to increasingly pay more for each byte of data delivered to them.

Why have the unit economics of the Internet changed so dramatically? “We built a road that was well-suited for bikes and cars and spent the money to build and maintain that more or less properly,” was the way one service provider executive explained it to me. “Now we have folks landing planes on the road, tearing it to shreds and making it unusable for others. So we need to spend lots more to maintain the road for bikes, cars and planes.”

Infrastructure technology like terabit routers, 60-gigabit backbone connections and multimegabit broadband connections do exist to support bikes, cars and planes — but the service providers have failed to spend the money from your Internet connection fees to invest in that infrastructure. Instead they have spent it supporting their bloated organizations and devising new pricing models to extract more money from consumers for less service delivery.

And therein lies the rub: The Internet has evolved and has enabled new applications such as peer-to-peer and video streaming that are increasingly being used by the consumer. Unfortunately, the infrastructure evolution of service providers like AT&T and Time Warner are working at a significantly slower pace. And that slower evolution costs them money, because their infrastructure cannot handle the new Internet applications, so instead of building efficient organizations that can evolve and deploy infrastructure faster they are looking for more money from the consumer in the form of usage-based pricing.

One day soon, when you get your Internet connection bill and it is much larger than you expected, don’t blame Hulu or Microsoft for offering you funny videos or a new security patch, blame your service provider for not evolving with the Internet.

I was recently talking to Richard Donaldson (an adviser of ours at Panorama Capital) of United Layer about a novel approach to optimizing data center cooling – using forward-looking infrared (FLIR) cameras.

United Layer rents a FLIR camera, he told me, the kind typically used to help pilots see at night or in dense fog, to create an infrared thermal image of equipment racks in which inefficient configurations can be easily detected. Once they’re found, United Layer works with the customer to redesign their rack layout, improving equipment performance, lifetime and total cost of ownership. Of course, this process also makes it easier to cool the data center, which helps control United Layer’s operational costs. As Donaldson explained to me in an email:

I think what we can see is that the “densifying” of the racks can become rather problematic when not thought thru – drive arrays should have spacing and be placed at base of cabs or close to cooling. I think that as existing facilities continually try to pack more into racks they will begin to really see how grossly inefficient the old “monolithic” paradigm of trying to cool the whole data center really is. The thermal images allow us to see exactly how each rack layout is good or bad for future design recommendations – all the temperature sensors in the world don’t give this kind of granularity…We also wanted to see real time thermal shots of our cold row heat containment strategies to prove and further illustrate what we knew from limited data.

Using FLIR to examine each equipment rack in a data center appears to be a novel and unique approach to optimizing cooling. I’ve seen thermal scans used to examine entire buildings for thermal leaks and hotspots, but never the use of FLIR to examine actual rack layouts in such detail. While a good quality FLIR camera sells for upwards of $10,000, Donaldson rented one for $80/day — what he called “an entrepreneurial approach.”

There are other new and interesting approaches to measuring the physical environments of data centers, such as the wireless sensors and benchmarking capabilities from Synapsense. But for $80/day I’m tempted to get Donaldson to hook me up with his FLIR camera rental company so I can check out our equipment racks myself.

Image credit: United Layer

When it comes to the operations of Internet businesses, 99.999 percent uptime, or five nines, is one of the critical metrics of reliability. Yet that metric — essentially the ability to say that users will reliably be able to reach a business’ web site 99.999 percent of the time — still eludes nearly every of them.

99.999 percent uptime for a web site equates to just 5.26 minutes of downtime per year. That is the total amount of downtime — planned or unplanned — as seen by users. According to a report last month by Pingdom, only three of the top 20 most popular web sites achieved this metric in 2007: Yahoo, AOL and Comcast’s site for high-speed Internet customers (eBay’s site was close, with only six minutes of downtime in 2007). Another report by Pingdom shows that most of the popular social networks did not achieve even three nines (or less than 525.6 minutes of downtime) in the first four months of 2008. Moreover, none achieved anywhere close to 99.999 percent uptime.

When I ask people about web sites that are down more than they would like, the most common response I hear is that the web site is a nice-to-have feature of their lives and not a critical element that they absolutely rely upon. If one search engine is down, you can use another. If a social network is down, then there are other ways of reaching your friends.

In other words, web sites are seen as unreliable, a perception that drives down user adoption, increases churn, reduces page views, limits ad impressions and increases abandoned shopping carts. I believe the converse to also be true — highly reliable web sites have high user loyalty and return rates, lower churn, more page views and advertising revenues and more sales.

While some may dispute the accuracy of Pingdom’s measurements, clearly being unreliable is not a trait that any person or web site strives to achieve. An unreliable person in life is not someone you invite home to meet your parents and not someone you want working on your critical business project. When interviewing candidates for jobs at my portfolio companies I have never heard anyone refer to their redeeming quality of being unreliable and absent more than expected — so why is that same quality so visible on web sites across the Internet?

It’s always better to be known as the reliable person in, any situation, whether for personal or business reasons. I wish that the Internet web sites that I use frequently in both aspects of my life had the same redeeming quality. Is 99.999 percent uptime too much to ask?