Last Thursday, after the Microsoft Hosting Summit wrapped up, the WHIR contingent had the opportunity to tour Microsoft's Redmond, Washington campus with lead marketing manager for hosted solutions Michael van Dijken.

The tour actually began at the company's Bellevue office, just a few steps from the Westin hotel, where the conference took place.

We didn't spend much time at the Bellevue office, but we did get a chance to check out the lunchroom on the top floor which has a beautiful (dare I say, belle) view of Seattle - not to mention the very strangely enclosed immediate vicinity in Bellevue.

Over at the Richmond campus, the really remarkable thing is not necessarily the offices themselves - which from what we saw were about what you'd expect from any normal office space - but the overall scale of the campus itself.

It's probably fair to say that with a few exceptions scattered throughout, Microsoft's campus takes up about just all of what we saw of the city of Redmond (like Bellevue, a sort of suburb of Seattle). Obviously that's not entirely the case, but according to van Dijken, Microsoft currently has in the vicinity of 40,000 employees working at the campus. And according to Wikipedia, a recent census puts the city's population at somewhere in the range of 46,000. So anyway, it's close.

Another remarkable aspect of the campus is the fairly excellent job Microsoft has done building around the green parts of the city. There are trees - really huge trees, as a matter of fact - just about everywhere you look. Nice place to visit and, presumably, a nice place to work.

We spent a good chunk of our time in the Visitor Center, which provided some entertainment, as well as some of the best photo opportunities. I suppose it's a testament to my vanity that among my favorites was a picture of me, but there were some pretty amusing scenes - and it's my blog, so I suppose you'll have to bear with me just this once.

Another of the really interesting features of the campus was the "Connect" system Microsoft has in place. Because it's such a big place, the company - in an interesting eco-friendly undertaking - has set up a system of hybrid cars to shuttle employees between offices on-campus, which saves them from having to get into their cars to drive a mile or two. It's worth noting the sheer number of these shuttles. They were really everywhere, making up a sizeable percentage of the cars we saw on the streets.

Finally, we set up with Michael van Dijken for a WHIRtv interview. The full video of which can be viewed on the WHIRtv blog page.

These are just a few of the pictures I took on the tour. There are dozens more you can check out on our WHIR flickr page.

I recently had the opportunity (actually, it was a couple of opportunities) to speak to Dave Roberts, vice president of strategy and marketing at Vyatta about the company's open-source networking platform. It's an interesting product, and as yet a bit of a departure for the hosting business in terms of the way these services are deployed

As usual, many of the key elements of the conversation made it into the article we posted on the site, but much more was said about the Vyatta community (it is open source software, after all) that helps to further fill out the picture of Vyatta the product. It seemed, as it often does, like excellent fodder for the blog

First, a lengthier explanation of the model Vyatta uses to package its open source product (Red Hat versus MySQL)

"Essentially, there are a variety of open source business models; what we typically call here, just for convenient shorthand, the MySQL model and the Red Hat model. In the MySQL model, a company writes all its code and then releases it as open source. So if you go to the MySQL website, you're basically getting code that's written 100 percent by MySQL with contributions from the outside world. If you go to Red Hat, you're essentially getting a distribution that is pulling components from all over. Red Hat incorporates MySQL, a Linux kernel, Apache, Perl and all these other components and pulls them all together in a distribution and gives you some hooks to be able to manage it. We are more like the Red Hat model than the MySQL model; we are pulling together components from other places, but we're also doing a lot of our own work to integrate those components."

This difference translates into a difference in the way the user interacts with the company's support, and with the product itself:

"Our fundamental belief is that Red Hat's community is different than MySQL's community because of how things are done. So if you discover a bug in MySQL, you're going to send it to MySQL, not to Red Hat, even though Red Hat incorporates MySQL. What you would submit to Red Hat are patches to Red Hat's specific stuff.

To a certain extent, our community is the sum total of all the communities of all the projects that we incorporate."

Finally, Roberts offered an explanation of who he envisions using the product, and how he sees that relationship contributing to the development of the Vyatta platform, both within the user's system and at the top level:

"We're designing for people who want to deal with our package exclusively, but we make it fully available for people that want to go deeper. Our philosophy is to present something that is very highly integrated and wrapped up, and is Vyatta, as opposed to just a collection of packages, even though we're not ashamed of the packages we use and we don't try to hide that. We'll tell you flat out what's in there

One of the things we see as an advantage is its based on Linux. So even though we're trying to provide a nice experience that is one of a traditional proprietary appliance in terms of being highly integrated - the fact that you can jump down to the Linux level, which we don't prevent you from doing at all, is an advantage to people. And you can extend the system, because it's based on Linux, with other Linux packages.

Essentially we're compatible with a Debian package format, and so you can add Debian packages to a Vyatta system. For instance, we don't ship it this way, but if you wanted to run MySQL on your router, you could. If you wanted to extend it with another package we don't yet provide, you could. You'll lose some integration when you do that. You're going to have to administer MySQL at the Linux level. You're not going to do that in the nice GUIs and shells that we provide.

One of the goals that we have over time is to allow the community to provide that level of integration for different packages that they find beneficial. And that's something that we see to a certain extent in the Red Hat community.

If I had to discuss the differences between the MySQL community and the Red Hat community - the MySql community is more of a developer community that is working on MySQL itself. The Red Hat community is more of an integrator community. So they're taking other packages that they find interesting for which red hat doesn't have any official distribution and basically making them red hat compatible."

I was talking to a friend this week about SWsoft’s recent move to buy up a few of the competing automation products in Ensim Pro, H-Sphere and Sphera, and he pointed me to an interesting note in SWsoft’s FAQ regarding H-Shere:

 

During the course of the transition, existing H-Sphere prices posted at http://www.psoft.net/price_list.html will be changed.

- H-Sphere prices will become consistent with HSPcomplete

- SiteStudio prices will become consistent with Sitebuilder

- CP+ prices will become consistent with HSPcomplete

- Support prices will become consistent with SWsoft support pricing

 

It seems at least a few people are wondering why and how SWsoft intends to change prices for H-Sphere products, and to what extent the company intends to continue its support for H-Sphere.

 

I happened to have SWsoft’s senior product manager Doug Johnson on the phone yesterday, so I raised the question with him. He said it was important to SWsoft that it not make any big changes to the business models of hosts using H-Sphere and any pricing changes would have more to do with normalizing support costs, eliminating excess SKUs and other similar fine-tuning type adjustments. 

 

 “We’re not going to do any widespread pricing changes. What it means is if there are things that are inconsistent about the pricing models, we’re going to make them consistent. So, for example, if support is priced one way on one product and a different way on a different product, we’re going to try to do some rationalization there. If licenses come in a package of five on one product and a package of 10 on another, we’re going to try to eliminate some SKUs and things along those lines. 

   

“In the short term we’re going to maintain two separate pricing lists so we don’t interfere with anybody’s existing business. And then I’m getting a lot of questions from forums and things along those lines about what’s going to happen five years from now, and I certainly can’t predict that. So we’ll just have to see how that goes.”

 

SWsoft has, a few times, pointed to Confixx as an example of how the company treats acquired technology with existing customers. The company bought Confixx in 2003, and has continued to support (and develop) the product since then, even issuing several new updated versions.

 

It sounds like H-Sphere users are quite safe. At the very least, they have a few years before they’ll need to consider migrating to another product.

Last week, I wrote a feature on the efforts of 3Tera, the company behind the AppLogic grid computing platform, and its efforts to extend the deployment of its technology beyond the service provider environment into the enterprise data center.

 

In aid of that particular article I spoke, as I sometimes do, to 3Tera’s VP of product management Bert Armijo. He is, needless to say, a man with a message, especially when the conversation has to do with the 3Tera products he manages. But he is also a reliably source of insight into his company’s business.

 

The direct-to-enterprise relationship is a new focus for 3Tera, which had previously aimed its product more squarely at the service provider market. As Armijo explains in the feature, it’s the same software. And as with the service provider relationships that came before, the relationships with enterprise customers require a certain degree of instruction in the platform and its deployment.

 

Excluded from the final feature, but worth putting out there, I felt, was Armijo’s description of the process - a fairly new institution, though roughly identical to the existing process - by which an enterprise would establish the relationship with 3Tera that would put AppLogic in their facilities, as well as some of the more common situations that accompany that process:

 

“What we have found thus far is that most of the enterprises we are dealing with actually have an initiative already internally, either to build a utility computing hosting service, or to look at utility computing services and how they can leverage them in their data centers.

 

“And so if this is an initiative they’ve kicked off already, they have people in charge of that or responsible for that effort. And when we engage with them, they already have some internal plan of how they want to actually leverage the technology. Their questions for us really are how does the system work? How does it fit into their data center operations? How does it fit into their scheme of software? 

 

“The process typically involves a number of face-to-face and WebEx meetings over the course of a few weeks, at which point they will take an evaluation system, usually hosted, from us and one of our provider partners. There will usually be two or three people involved in that application and they will pick a test application and migrate it over to the hosted grid. That will be the basis for tests that they will run in terms of how does the system operate? What does it take to operate it? How does it deal with typical hardware failures or human error? 

 

“This is where we get into customers requesting licenses of us. They then make a decision as to whether they want to proceed with hosted resources or with internal resources. And if they want to proceed with internal resources, their next step is to get a license and build their first grid internally, which we help them with. 

 

“They will pick a few more applications and start moving them onto the grid. This might happen one application per month, or one application per week. Or we have one particular customer who is moving a couple of applications per week on to the grid. 

 

“During this evaluation stage, they’ll look at the operational aspects of the system, start writing their policies and procedures for how to actually manage the grid and how to manage the applications on the grid. This evaluation process can be anywhere from a couple of months to six months depending upon the company and the complexity of what they’re trying to do.”

 

Yesterday we posted a feature about The Planet’s new plan for distributing a hosted version of Microsoft’s Small Business Server.

 

Last week, I spoke with Urvish Vashi, director of product management at the company.

 

As is most often the case, there was much more discussed in the interview than could possibly fit the feature. And as has become somewhat customary, I think this would be a good place to let a few more of those ideas loose.

 

One of the most interesting aspects of the announcement is the decision by The Planet to pursue not just a new market, but also a new channel for reaching that market. Looking to spread its hosted Small Business Server offering throughout the small business market required the crafting of a partner program for the systems integrators who quite often do the IT decision making in that space.

 

For much of the conversation, Vashi discussed the specific value of the service to that systems integrator community. From the perspective of the hosting business, and most likely from the perspective of the small business, the value of a hosted solution is evident. But for the partners The Planet is looking to acquire, a considerable shift is required in deployment model, pricing and the way they approach their business.

 

Vashi says, probably rightly so, those small business specialists have a lot to gain from pursuing the switch, turning their static relationships into sources of recurring revenue.

 

For The Planet, the most interesting aspect of these prospective partners is their ability to customize the solution for the specific niches with which they most closely associate:

“I think this kind of keys in on the model we selected. These small business specialists often specialize in niches that I would never have thought of. There’s the guy that’s doing legal in Atlanta. There’s the person that’s doing non-profits in DC. They all have very different needs, and often different applications. Unlike some other markets, where you might be going into a cookie-cutter thing with shared Exchange or SharePoint, we decided to go with a dedicated offering, which means that the offering is a dedicated server for each individual end customer. So that partner can now install, configure and manage whatever third-party application they might have. Maybe it’s a customer homegrown application.”

The Planet has every intention of cultivating the channel for future growth. But the task at hand is creating a profitable relationship with the resellers while letting them steer the already-profitable relationships they have with end customers.

 “I think we’re going to learn a lot from these customers. And as we cultivate this channel, I think they’re going to take us to the next set of products. Maybe it’s going to be something like Microsoft Dynamics after this. Maybe it’ll be something like another CRM package, or financials.

“Once we create a profitable relationship with these partners – they’ve already figured out how to have a profitable relationship with their customers – we’re going to let them drag us any way that makes sense.”

It has become a fairly regular occurrence that I follow up a feature with the sort of "deleted scenes" blog post detailing some of the discussion from the interview that might not have been fully represented in a feature.

Quite often, an interview will stretch over half an hour, or longer, and almost invariably, interesting things will be discussed that don't fit the final cut of a given story (the actual volume of this material varies from interview to interview, of course, depending on the pace of the conversation and how pointed it is).

Last week I spoke to several executives from Absolute Performance Inc., a developer and provider of performance monitoring technology that might be of interest to a hosting provider either as a tool to be used or as a product to be sold.

According to John Galloway, VP of Sales:

"About half of our business comes through enabling hosting providers, partners who essentially white-label what we do. The other half of our business is actually helping customers directly. So obviously the channel side of the business that we refer to, we're helping those partners. A lot of those are infrastructure-providing hosters. There's a percentage of them who are software providers. So they actually embed some of the things that we do into the software solutions they provide their customers. There are even more classic outsourcers or consultants that we work with to help provide a solution for enterprise also."

He says one of the popular uses of API's monitoring technology is among application providers who want to provide a more intricate means of support.

"The classic example would be a workflow application where the workflow provider has a lot of exposure to tenancy on external applications - for example, if one of the steps is to send an email to somebody. If the email system is down or is having problems or is slow, and the client isn't directly aware of that, suddenly the workflow system is thought to be the bottleneck, and people on the floor aren't able to do their tasks and queues back up and it becomes a problem. Our ability to link or extend outside this workflow process as well as understand what's going on on the gear that's running the application is kind of where the value proposition starts."

Part of the program at API is deploying the infrastructure to support its monitoring solution in a partner's data center (no small investment, at roughly $250,000), which enables that partner to deliver privately branded monitoring services from within its own facility.

Galloway described in some detail the company's relationship with a specific service provider (although he asked that the company not be identified directly; apparently they hadn't discussed the matter of publicizing the relationship.)

"They're using our monitoring platform to better monitor their internal control systems for provisioning and billing, and they're getting ready to roll out a managed exchange solution with a client. So they're going to provide that software as a service, and they're going to use our tools to monitor that. In addition, they're using our capabilities to monitor, at a basic level, all the customers that come into their shop - they have some obligations around network availability and those kinds of things that they do as part of the basic services. Then they're in the process of launching a kind of a scaled down OS monitoring solution as a first up-sell. The second-tier up-sell is around exchange, oracle, apache, IIS, Tomcat,  that kind of standard list of application infrastructure products.

"At the higher level, where the customer actually has a Web site they're running or an ERP application or whatever it happens to be, we would be directly engaged with those clients around managing the database, or monitoring end user experience around the web site.

"So it's everything from helping them make sure their lights are on at the data center to helping clients, almost on a consultative basis, be more effective with the systems they have in their shop."

I spoke to Bert Armijo from utility computing platform developer 3Tera this week, mostly for a feature to be published next week. But some interesting things came up that aren't perfectly suited to the regular feature, and I wanted to get them up via the blog.

A bit of preamble: 3Tera's business model isn't totally typical of a supplier to Web hosts. It's not dependent on a standard licensing fee, and it's a piece of equipment at a set price. 3Tera's income is based on the amount of resources being deployed under its AppLogic platform. And its involvement in the distribution of the service is integral enough that the word "partner" is actually, in this case, a very accurate description of the relationship between Web host and vendor.

So, while 3Tera's goal is certainly to distribute AppLogic-based utility computing to the largest number of end users possible, that doesn't necessarily mean distributing it via the largest number of Web hosts possible. In fact, says Armijo, the company is most interested, as of this moment, in working with a relatively small number of partners, and focusing on making them as successful as possible.

The result of all this is that 3Tera's major objective is not to sell Web hosting providers on the value of the utility computing model; its objective is to sell end users, would-be colocation customers, on the potential of the utility model in general, and the AppLogic platform in particular.

Preamble complete.

Armijo is excited about the growth 3Tera is seeing in the audience for utility computing services:

"Last quarter we saw more than 100 percent growth in the number of customers using virtual private data centers. So it was a very successful quarter. And we're seeing that growth continue into this quarter.

"We're seeing a pattern of users that basically are those folks that we had been targeting since the middle of last year. SaaS and Web 2.0 providers, primarily. And a few of them are kind of prototypical of the business drivers that we've been talking about for a while."

Interestingly, he says that a big part of the interest can be attributed to Amazon's Elastic Compute Cloud service - the very large Amazon brand making the case for utility computing in a way and on a scale that is simply beyond the means of the comparatively tiny 3Tera.

"To be quite honest, the guys at Amazon did us an enormous favor in launching last year because their visibility has really pushed forward people thinking about utility computing. And in the research people do before they sign up a lot of them do find us, and our partners. So it has really helped us to start the conversation with a lot of companies that otherwise might have pushed back a little bit."

[Case in point: while writing this, I Googled "Amazon EC2," and was met, among other things, by a sponsored link to 3Tera's site that asked if I was "waiting for EC2."]

Amazon, says Armijo, is not a company that 3Tera considers a competitor. This is because Amazon's role in utility computing is as the distributor of a service, whereas 3Tera is the developer of an underlying technology. Conceivably, Amazon is a company that 3Tera would consider a possible customer. However, 3Tera's partners undoubtedly consider Amazon a competitor at this point.

But he says the services do deliver two very different varieties of value.

"The two systems are different enough that there's very seldom any difficulty positioning and getting customers to understand the value proposition of the two. AppLogic is all about systems. It's all about large-scale infrastructure that you can actually provision load balancers and VPNs and firewalls and things that don't make a whole lot of sense in the Amazon Web services environment. On the other hand, Amazon is all about resources by the hour, and if what you want to do is fire up a lot of virtual machines, you can do that in a very quick, easy way."

The point being that Amazon, with its very audible trumpeting of the model, is doing a considerable share of the work with regard to 3Tera's objective of selling the idea of utility computing to potential customers.

This week, I spoke to Scott Cutler of anti-spam and email security company AppRiver - a fact mentioned in my previous blog entry, and alluded to by his repeated reference in a feature posted this afternoon.

It was a very long and interesting interview, and it included a much broader discussion of the world of spam than could fit into a feature, or the tail end of a blog posting on SSL certificates, and I thought this might be a good place to expose a couple of interesting points from that interview that deserved to be seen, but didn't make their way into either of the afformentioned items.

In particular, we discussed botnet technology and its effectiveness in foiling the efforts of both those who would block spam (such as AppRiver), and those who would identify the spammers themselves.

The effort to build botnets these days is so intense that virus writers build controls into their programs that scan a newly infected machine for the presence of competing viruses, and wipe them away before installing new malicious code.

The more interesting aspect of the botnets, however, is that they distribute the commands and controls in such a way as to make it impossible to trace the source of those commands.

(Admittedly, the nuts and bolts of this begin to get beyond me.)

According to Cutler:

"The risk for the botnet creators is that if I were to capture one of these PCs that was infected and look at the code, I would be able to decipher where those instructions were coming from, and I could go upstream from there and shut down the whole network.

But they have a cellular structure to them. So the actual bot on the PC, it can look upstream to a number of places that can give it the instructions as to the next spam it's going to send out. Let's say it's grabbing that from a Web site in some Web hosting company in China. And it's been reliably doing that for a few days. If the Web hosting company in China discovers that one of the servers in its network has been compromised by one of these botnet management [groups], and the hosting company discovers this and shuts the server down, the botnet is smart enough to go find another source in order to continue doing its work. And it constantly updates the places it can go look, but it doesn't have the whole variety.

So apparently, the way the structure is built is that you could grab any one, or any 10, of these PCs that are part of the wider botnet, but you could never find all of the PCs and you could never find all of the command and controls upstream from them. So trying to shut down one of these botnets is amazingly hard. They're extremely fault tolerant."

Interestingly, the gravity of the spammers' control over the botnet situation and the possibility of unraveling or reversing that scheme was revealed by what was supposed to be a breakthrough in fighting spam.

(Granted, this is sort of speculative hearsay type talk, but think of it as a folk take if it helps you. It's illustrative of a more significant point.)

"I heard, and I don't remember where I heard it, there was a group of folks that was trying to unpack the botnet. And they were pretty excited about it. They thought they were going to figure it out. And when they peeled the onion back one more layer, they figured out that they weren't even close, and they're never going to be able to figure out how the botnet works.

And the group that had said that - this is a little while ago now - said the fact of the matter is, after our whole research project, we think the spammers are probably three to five years ahead of the anti-spammers.

And my personal view is that that's probably true."

Earlier today I spoke to Gabriel Murphy, who recently acquired - and became CEO of - Web hosting provider Aplus.Net. The interview will appear in the upcoming issue of the WHIR magazine (and eventually on the Web site), but there were a few interesting points I thought it would be worth sharing right away.

As you may know, Murphy founded Communitech many years ago, and at one point owned iNet Interactive, which operated HostingTech and WebhostingTalk.

We talked about his recent investment in Aplus.Net, and his plans for making that company into a force, in terms of technology and scale, in the Web hosting business.

A big part of that plan to add technology and scale involves acquisition. And he hinted at some upcoming news to that effect.

Murphy and Catalyst Investors bought Aplus.Net quietly in July of 2006, but didn't officially announce the deal until December. Part of the reason for that, he says, was that they were waiting to announce another acquisition at the same time.

He said the new deal is about 45 days away. And about that deal, he said this:

"We're not buying $5 million to $10 million hosting companies. Were looking to scale. We're looking at substantial Web hosting firms in terms of size."

You heard it here first (I think): expect Murphy and Catalyst to announce another "substantial" acquisition about a month from now.

Tags: gabriel murphy, aplus.net