You are the reason vendors making appliances are suddenly treated not like misdemeanor crooks stealing business from the big boys of the server businesses, but as serious rivals. You are the reason yesterday’s startups making appliance servers get bought.

We probably won’t be breaking any news by informing you that Cobalt Networks, a company that made a name out of selling appliance servers, is being acquired by Sun Microsystems for $2 billion. The transaction, subject to various regulations, is supposed to close this month.

Sure, we have heard explanations that Sun is trying to get into the low-end hosting market normally dominated by Microsoft, and that it couldn’t afford to spend time building a brand for its own line of products in this space. But dry numbers and analyst-speak describing this deal really don’t do justice to what is taking place right before our eyes.

In the time of telecom companies seeing their shares being flushed down the toilet, money keeps flowing into data centers. Everybody and his mother is setting up web sites. Housewives take up website development and launch web hosting firms on the side.

Web farms pop up left and right that host mostly other web hosts, the virtual kind. The kind that would buy a dedicated server, but wouldn’t know what an Ethernet jack looks like. Welcome to the latest revolution. Welcome to the wondrous appliance-land.

IT'S A BLUE WORLD
Before we go any further, let’s set the record straight as to what is an appliance in the web hosting market. Most likely, it’s a Cobalt box. Sure, there are other vendors out there. Intel has a product out that kinda looks like Cobalt. Network Engines has a box on the market as well.

But if you are into reading the trade press — rags like ours, only pushing more legacy-type equipment for other industries — you are in for a twilight zone-kind of experience.

“The main players in this space,” says Michael Gilpin, an analyst specializing in appliances at Giga Information Group, “are IBM, BEA, iPlanet, ATG and even, to some extent, Silver Stream.”

Don’t panic yet — these are not likely to be devices massively used by web hosts. Enterprise customers have the same set of issues as carriers, and the same product features that make appliances click in a data center make them indispensable in an office: Easy to set up. No maintenance. No special skills to run.

In fact, Cobalt’s first entry into this market was with the Qube product — an appliance for small business, designed to be an on-the-premise web and mail server. Of course, then a couple of ISPs had a talk with a couple of Cobalt guys, and they saw the light. No one did that favor for IBM, or for Gilpin, for that matter. He says it best himself: “While we see a large variety of companies, most often we talk to enterprises buying technology for their own use.”

So, unless web hosts prefer to pay big bucks to Giga to ask Gilpin questions about how to do their work better, it is safe to say the vendors he has mentioned are the ones used mainly by traditional businesses seeking simplicity in life by delegating painfully difficult to manage services to appliances. By buying expensive subscriptions to Giga research. Nothing wrong with that, just a different market.

Back on the farm — sever farm, that is — Cobalt rules the roost. The vendor was among the first to help services providers — ISPs and web hosts — empower their customers to self-provision their services. This simple concept went a long way.

A common perception in the vendor space for years has been that carrier-class equipment has to be feature-rich and complex, so that carriers — experts in what they do — could fine-tune it to their intricate requirements. On the opposite side, enterprise equipment had to be very simple, so that technologically inept business people had fewer chances to break it. Copy machines are a good example. Xerox is still capable of selling a lot of businesses on its maintenance plan, when a Xerox man comes to the premises to fix whatever is wrong with the copier.

Web hosting presented an interesting dilemma for a company like Cobalt. If there were a server that could not only be easy to install and configure, but also could be managed with almost no brainpower, why would a carrier want it?

It’s the economy, stupid.

No set up cost, no maintenance, no customer support, no provisioning. Customers do all the maintenance work themselves, and they don’t need to know anything about networking in order to run their web hosting businesses.

“I don’t think that most of our customers know that they are using Linux,” said Bill Mann, president of Host2Own, a San Diego-based web host with about 700 Cobalt servers on its racks.

And then there is the cost of the Cobalt server. Host2Own resells it for $250 per month. For an individual willing to sell web hosting services, this means a break-even point at nine web sites, given that a service retails for around $30 a month.

Now, way before it was popular, Cobalt created proprietary software that enables virtual server creation. Each Cobalt rack server therefore could support 250 web sites.

No wonder web hosts line up their racks with Cobalts, and then have resellers launch “virtual” web hosts by leasing as little as one server. While all that we have talked about so far is mostly circa early 1999, don’t let anybody tell you the Cobalt opportunity is old news. The time to sell this stuff is now.

“In three months, we want to become a diamond partner with Cobalt — which means we are in the million-dollar range,” said Mark Bayliss, president and chief executive of a web host with the telling name CobaltRacks.com. The enterprise wants to become the largest appliance-centric web host by early next year by claiming to be the first single operator to have 1,500 Cobalt servers online.

COMPETITION NEVER SLEEPS
You’d have to figure that with the number of academic types in the Internet industry not diminishing over time, somebody would have figured out by now that Cobalt is onto something, and would have tried to replicate the success. Good assumption.

Starbox, NetMachines, Network Engines, Intel’s Net Structure, CrystalPC, Siliconrax and ServeLinux are all among the newcomers to the appliances space. It’s probably fair to say these players as a group have a lot to prove to web hosts, especially those who have become over time Cobalt bigots.

“So far we have tried only Network Engines, and it isn’t even close to Cobalt,” said Bayliss, “… to be honest with you, our own code put on a white box works better.”

There is also a way to build your own appliances. A variety of software companies are offering virtualization software that in its first reiteration lets users parcel any server into hundreds of virtual servers, and in later versions offers hook ups into more sophisticated applications and gives turnkey management of back end-type functionalities. In plain English, it lets you built a set of Cobalt-type functionalities on a server platform of your choice, with applications of your picking.

In the running here is the 800-pound gorilla Ensim, rapidly closing in on Sphera, and smaller players like Plesk (which, albeit small, has a remarkable market penetration), H-Sphere and Systemsfusion. Then, there are newcomers like Professo; and there are even rumors that Alabanza, a shared server-farm powerhouse, might make its automation software available as a standalone product.

Not surprisingly, the two schools of thought are clashing. Cobalt-type solution advocates say an appliance is an appliance when all the pieces that were meant to work together have been tested, and do in fact run smoothly on existing hardware. Ensim-type solution supporters argue for flexibility and ultimate cost reduction with unlimited ability to add new applications. They also point out that investment in existing server farms is not trivial and should be enhanced with software, not scrapped by buying all new hardware.

“Our technology enables making [the] provisioning of a server farm as easy as drag and drop, and additional applications can be added just as easily,” said Ensim President and Chief Executive Ken Fehrnstrom. “And with customer churn being an issue with some of the bigger web hosts, we can help these operators curb customer dissatisfaction.”

The next big fight among vendors is to figure out how services hatched on an appliance’s platform would develop.

INNOVATION AND BEYOND
It might come as a surprise to some, but one of the top selling technical books in Japan is NTT MMCube. The technical manual describes how to connect a Cobalt Qube to a cellular carrier’s internal network, so Japanese mobile telephony providers can seamlessly roll out wireless email. MMCube is a Cobalt white box, specifically altered by Japan’s incumbent carrier NTT for the wireless email functionality.

The book’s success goes to show that, with more than 200,000 independent developers out there inventing different applications to ride over popular appliances, innovation is anything but stale with both Cobalt-like and Ensim-like solutions.

Is this sophistication lost on virtual web hosts who have just mastered the art of provisioning virtual servers? That is the question.

On one hand, operators of web farms see unsophisticated web hosts paying little interest to any other features that exist on a Cobalt platform other than provisioning. On the other hand, the technically savvy part of the bunch that knows to want firewalls, e-commerce, outsourced billing — the works — is very price-conscious.

“I’d say that if they could save $200 by building a new feature themselves, they would probably go ahead and do it,” says Bayliss.

The gamble that Cobalt and other vendors are making — which, right now, looks like a sure thing — is that unsophisticated web hosts who are interested only in easy provisioning today would want to sell more features to their customers tomorrow. The logic here is that with enough customers asking for splashy web sites with e-commerce and security features, the general public would know to ask this question to web farm operators.

And that, folks, is where marketing hype turns into a marketing tornado. You’ve got appliances that do nothing but firewalls. Nothing but billing. Nothing but email.

“We see our audience as global 1 million — the worldwide population of small businesses,” says Peter Ulander, Cobalt senior director of product marketing. “These are the people [who] have not been in this game before, and they are just now getting to put their company online.”

The market is coming at this from the other side. Look at this Japanese Cobalt implementation — we guarantee their business development people have not seen it coming. At the latest trade show, Cobalt had half a dozen partners just in its booth demonstrating various bells and whistles added to the platform.

“We let you set up billing for up to 200 stores on one server,” says Curtis Pierce, president and chief executive of Kurant, a billing-platform provider. Pierce was most excited about the latest functionality of his platform, where a web host would be able to access its billing database via a wireless browser.

Now this is an image we could relate to. You are on the beach in Timbuktu, spending hard-earned cash on various rum and juice blend beverages. You pull out your cell phone, as it is the 15th of the month, and quickly scroll though the list of customers that have not paid. You pick up a couple that are behind and call them to say that they are about to be turned off. You can do that, you know, right there, on your cell phone, on the beach.

Some might say that taking your business with you as far as Timbuktu is insanity. True. Because you can doesn’t mean you will. The point is that these are the kind of applications vendors are thinking of to help web farm operators sell more services to their clientele. Who knows what they would buy and what they wouldn’t?

“I think the old web hosting model where you have to pay to host is dying. Soon everything — bandwidth, colocation, server — [will] be free,” said Host2Own’s Mann.

“What you will charge for would be the service — and managing the server for a customer is just a start.”