For those of you who get the reference in the title, rock on. For those of you who don't, write me off as a total nerd who reads too much. It's cool.

Yesterday was supposed to be a PTO day, but one of my co-workers pointed out a response that had been made to my "Facebook is not IT, But is it Cloud?" post from last week and it got me thinking. Derrick Harris, over at GigaOM did a good job of expanding on my original premise (thank you Derrick!) and points out that I may be ignoring the overall trend of companies/applications moving to the cloud at scale. The more I thought about it, the more I think he's correct, I am downplaying that aspect. But I'm doing so willingly: I don't think that's where the action is going to be long term. It always comes back to the Service Providers. I think the best way to illustrate my point is to look back at where Service Providers came from. You see, we've seen this before…..

If, in the early 1990s you had looked at the list of the top 500 data centers operators in the world by size, my guess would be that large enterprises would dominate the landscape. You could make an argument for telcos as well, but since we are talking specifically about compute resources here, we'll leave them alone for the time being. Don't worry, like all good things they come back around as well…

These enterprises are right in the thick of trying to figure out how to integrate their legacy mainframe-based systems with this new "internet" thing, but even the concept of client-server was fairly new. The "internet" (remember, the first graphical browser didn't come out until 1993) was still something relegated to the few, and to the educational environments. The specific technologies and the specific time-line aren't what are important, though…

Encarta has a very interesting definition for "commoditization":

making of product into commodity: the process by which a product reaches a point in its development where one brand has no features that differentiate it from other brands, and consumers buy on price alone

In technology, almost everything is driven by this process in some way. I look at it as the process by which we get all the nonsense out of the way and focus on scale and futures. There was a lot of angst around processor technology 5 years ago. So many choices, what is an IT shop to do? SPARC, PowerPC, (ahem) Itanium, AMD, Intel… Chuck Hollis from EMC wrote an entire article about the commoditization of this space, asking if the game was over in the CPU wars. The truth is that Intel has long been the dominant player in that space, but that they "won" is as much a comment on how their product has been positioned as it is on the product itself. Virtualization made a commodity of CPUs, with seemingly everyone but Oracle standardizing on x86. Intel fought hard to make x86 processors the de-facto standard.

That's not to say that vendors don't fight hard against this process. On the contrary, some work hard to stem the tide. VMware fights hard to keep virtualization from being a commodity, working to consistently innovate in the face of competitors like Microsoft, who realize that if this space becomes a commodity they have a good chance of becoming the winner due to their market position and size. And really smart companies like VMware have lots of irons in the fire, both planning for and delaying commoditization. You didn't think all those investments in Spring and vFabric were done for no reason, do you? If the hypervisor becomes a commodity, the next battlefield will be the application development and execution space, something that VMware, Microsoft and others recognize, and something that Oracle has been involved with for some time…

But back to the original point: we aren't looking at those large enterprises who dominated the compute-based data centers in the early 1990s as the leaders in the data center space today. Why: Because the data center became a commoditized business. I'm not being overly dramatic when I say that I could give $7M to almost anyone reading this blog and with a few phone calls and liberal use of Google you could build a data center. It's all round pegs and round holes today, unlike 20 years ago where most everything was adapted from some other use or custom-built. Once that happened, most enterprises realized that their involvement in this line of business wasn't providing any positive return to the business, and thus began the data center hosting business. Of course the largest of those companies had too much invested to just walk away, and even if they wanted to, they were too large or had requirements that an outsourced data center would struggle to meet. Maybe 200 of those top 500 sites had to stay in the data center business, but the rest now had an alternative. The big winners were the smaller companies who couldn't afford to build and operate their own data centers, because now they could get the same benefits as the larger enterprises without the capital costs. If you absolutely must, you can think of this as "Cloud 0.1"…

Now, let's do something cool. Let's take the top 500 Alexa sites and use them in the story above instead of the 500 largest enterprises. Right now, most data center outsourcing providers are probably completely unable to service the needs of Google.com, Facebook.com, YouTube.com and Yahoo.com, which is why those companies are building their own data centers. As Derrick rightly points out, it's not just the facilities they need, but it's the way that the compute and storage scale out that's important as well, hence the "open sourcing" of the servers themselves. Dell and HP and Cisco and others would be smart to start listening to this sector more intently.

If I had a crystal ball that looked into the future I'd have a lot more money and a nicer car. What I do have is the ability to see that what goes around comes around. If I had to hazard a guess, I'd say that the proliferation of web apps that run at scale is already driving an evolution in the hosting market to address that need. Why do I say that? Ask yourself this question: is Facebook owning, building and operating a data center core to their business? Of course it's not. The fact is that they had nowhere else to go, much like those large enterprises back in 1990. Just because it's a different kind of workload being run in the data center at a different scale doesn't mean history won't repeat itself. The first generation of these data centers already exist for the traditional retail sites, with the likes of CraigsList, Technorati, LiveJournal, TypePad, AdBrite, the 1Up gaming network, Second Life and Yelp all using outsourced data centers as far back as 2007. In fact, Digital Realty Trust claims that it provides data center solutions for over 10% of the Fortune 200 today.  One step down from the actual hosting of applications, the development of those new apps is also taking place “in the cloud” with notable examples like Dropbox and today’s Cloud Foundry announcement by VMware.  This definitely won’t be the last time this particular wheel spins around!

So, no, I don't think I'm ignorant of the proliferation of web-based apps and the impact they are having on application use in general as Derrick has charged (plus, I hate seeing my name and the word "ignorant" in the same sentence…). I recognize it, I just don't think that it's going to change the fundamental way of things; large companies go first, technology gets commoditized and then the service providers step in and fill the need for customers across a broad spectrum. We've seen this happen before, and Ka is a wheel who only knows how to roll…

(FYI: I'm not at all disparaging the talents and efforts of all the data center engineers out there. As a reformed facilities director who oversaw ~60,000 sq/ft of data center space in Charlotte, I know how hard that job can be. Just because everything is designed for the data center doesn't mean that it always works the way it's supposed to. They work hard in a dangerous environment and no one notices until something goes wrong. If you know a facilities engineer, especially at a multi-tenant data center, give them a high-five, or maybe buy them lunch. They certainly deserve it.)

 

23,410 total views, no views today

 

Comments are closed.