Suddenly, infrastructure is cool again

About two years ago, Jeff Bezos used the occasion of his appearance at the Web 2.0 Summit to talk up the merits of EC2 and S3, Amazon's entries into the then-nascent area of cloud computing.

The predictably perky CEO was enthusiastically regaling a standing-room-only ballroom about a future in which his company would sell data storage infrastructure and server capacity by subscription: the idea being that customers of the new services could move quickly from idea conception to a successful product by farming out the infrastructure side for Web scale computing to Amazon.

Now, that's hardware!

(Credit: John Deere)

"We make muck so you don't have to," Bezos joked.

It was an interesting idea but a left turn for Amazon. I remember that the guys sitting next to me weren't equally impressed by the pitch. They joked among themselves that Amazon was a company that sold books. Who was Bezos kidding?

Well, we know how that story turned out. Bezos' timing was propitious. Amazon happened to go into Web-based services around the same time that customers had started to lose their fear of "the cloud." Not everyone, mind you. But increasing numbers of start-ups and small companies were receptive to the idea that they could increase their server and storage capacity on a subscription basis. With millions of people going online to store data, run applications, or communicate via Webmail services--and all that functionality stored on the cloud--now they could participate in that computing shift without breaking their budgets.

This all was proceeding apace. But it was a slow transition. Then came last fall's financial meltdown, and suddenly, cloud computing's proponents had a timely marketing message. On the "Charlie Rose" show Thursday night, tech pioneer-turned-prescient-doomsayer Mark Andreessen, took note of the impact this computing shift has had on the tech business.

"So you've got a whole generation of start-ups that are basically just a couple of programmers with a couple of laptops, and they upload everything into the Amazon cloud. It's pay-by-the-drink like utility. So all of a sudden, you have this whole new wave of Internet start-ups getting started for practically no money, right? So there is a level of innovation. Every kid coming out of Harvard, every kid coming out of school now thinks he can be the next Mark Zuckerberg, and with these new technologies like cloud computing, he actually has a shot."

That's the classic sales pitch on behalf of cloud computing but Andreessen basically has it right. What's new is that with the economy going through a rough patch, this is turning out to be one of the few bright spots in an otherwise gloomy tech landscape. IDC predicts that cloud computing will account for about 30 percent of new growth in the Internet over the next three years. Poor economy or not. "It's not a surprise," said Frank Gens, the firm's research chief, who contends that the financial crisis only magnifies the benefits of the economics behind this computing model.

But here's the rub: While many executives responsible for their companies' IT operations grok the vision, they still refuse to make the switch. More than 60 percent of the companies surveyed recently by Kelton Research reported they did not use cloud-computing technologies, and most of them have no plans to use them anytime soon.

Chalk up their lingering resistance to a couple of old bugaboos which have been around since the days when "MIS directors," pressed to decentralize their computer operations, ruled the tech roost: security and fear of loss of control. If past is prologue, those issues will get sorted out over time--the same way that the sundry issues surrounding client-server and Internet-based computing models ultimately got resolved.

Capacity on demand is a big deal, especially for start-ups that are in no position to be buying servers. With API deployment, you say "launch" on these virtual servers and you're off to the races. It's just a lot more convenient to have an Internet-based data center.

Not long ago, Sun Microsystems' Dave Douglas told me that "every one" of his conversations with customers ultimately comes around to a discussion of where the cloud is heading. That pretty much jibes with what I've heard from executives at other hardware and software companies. Pay attention to this trend because it's taking place in real time, away from the media's glare. In an age where Twitter, Facebook, and a laundry list of forgettable social network doo-dads have dominated our attention, all of a sudden, infrastructure has become cool again.

Who woulda thunk it?

I like CNet, but this is another example of a writer that doesn't understand the business about which he is writing. Loss of control and "old bugaboos" are perhaps some small part of the slow adoption among IT folks. Those who "know" keep telling us that online services are the wave of the future (remember the "ASP" model? By marketing "magic" it is now rebranded as SAAS and the Cloud, hello?!?!?!), well it will continue to be the wave of the future until prices come WAY down, bandwidth goes WAY up, and security models are integrated with the clients' environment (i.e. with AD).

This is not to say that some software, well-suited to the environment, such as salesforce.com, etc will see great gains, they will. Progress will be incremental. I see great things for AWS, but it won't start until they cut prices about half. Right now their cheapest Windows server is still slower and more expensive than a dedicated server available from any of a dozen reputable providers. In addition, the client side just doesn't have enough bandwidth and good integration with AD is lacking.

[charlie cooper]

wow, thanks for the gratuitous smack in the head. and good morning to you, fella. actually, i do think i understand "the business about which he is writing." Let's separate out fact from fiction. I'm not saying this is going to effect a lock, stock & barrel transformation. And sure, there are legit reasons to go slow. But levitate 30,000 feet and take a look at the bigger picture. A lot of IT decision makers are choosing cloud because they like the economics, especially in a depression. (And yeah, this is a depression.) I've heard variations on your complaint about previous shifts in computing. It's the usual refrain and not very insightful about where things are heading. Prices do come down and performance - in this case, bandwidth - will improve. Etc. etc.

[rapier1]

As someone that deal with iT infrastructure (backbone carrier networking) all I have to say is that a lot of IT managers are going to end up being disappointed when their DRMPs don't take into account a backhoe or a failed transponder.

[geolemon]

They aren't "old bugaboos" - they are legitimate concerns that would have to be talking points in projects regardless of storage paradigm.

There *are* more opportunities for failure modes, for security breaches, even if they *have* all been handled. You might consider security has been handled under some sort of paradigm as well, which means a hacker who understands that paradigm could breach it - publish it if he's a good guy, exploit it if he's a bad guy.

And as appealilng as having redundant storage on the web is (I like having my backups done to Mozy, with their redundant servers spread out over North America), there's still a risk - the possibility of something more central/fundamental breaking, or the third-party company going out of business.

It's something worth discussing the pros and cons on - I hope the intent of the article wasn't to imply that this is always a superior solution. There's pros, and there's cons.

[conquer4]

So here is a problem i have been having, and maybe you can shed some insight on it for me cooper, esp with the rise of netbooks lately. The 'cloud' is a nice concept and place, although geolemon points out correctly that there are more security issues/problems. The thing that befuddles me is simply, connection. Most ISPs are implementing data caps, it varies from the ok (comcast-250GB/mo), to the bad (time warner-40gb). I realize this doesn't much apply to google or amazon or microsoft, who can buy whatever bandwidth they need, but for us consumers, i get somewhat wary when the entire point of my netbook could be pointless if i went over, and all access to my data effectively gone (unless i go somewhere else). ...Its like having a computer that i can't access over ~150gb of data on the HD for a month or it'll shutdown and be as worthless as a brick.
So am i missing something here? to me netbooks and online/downloadable HD movies seem worth less then the 'wave of the future' i see given to them in light of data restrictions.

[charlesneville]

What would make sense for the Content Delivery Networks would be to have edge points inside the ISPs networks, for the most popular files they host, then they are effectively local caches and the customer shouldn't take a hit on their bandwidth for files.

Come to think of it, why aren't ISPs partnering with trusted online backup companies like Mozy to provide the backups within their own network as a value-add to their existing customers, following the exact same model as free local phone calls?

The bandwidth caps are there to stop 'excessive usage' and because ISPs are paying for their bandwidth on a meter. This would neatly side-step that by using bandwidth more efficiently AND make things faster.

[Reader11]

The cloud is an interesting and necessary evalution to commoditize expensive infrastructure. tools will be required to make this model as successful as centralized computing and distributed computing. From storage-as-a-service, monitoring-as-a-service, security-as-a-service availability and cost are core drivers successful innovation, that as-a-service tools provide. These tools booked in accounting as an operational expense as you only pay for what you actually use, directly associated with the application or transaction of the end customer in most cases. Agree with the author, the current economic distress will push the cloud computing ( infrastructure-as-a-service ) model through the door which ASP wasent able to achive.

[inverse137]

If you had been able to work in 'synergy,' you would have covered every major industry buzz word. I hate marketing speak.

"Necessary evalution to commoditize," are you kidding me? Who talks that way?

MBAs and marketing majors should be beaten.

[alainyap_morph]

I hope not everyone is buying into the logic that cloud solutions offer no value just because even an innovative company like Coghead suffers a mishap. On the customer side, this is a good example of the perils of a proprietary solution. It may be really cool but then business model fails and down goes the apps. Another thing worth noting is that unless 'enterprise adoption' of cloud solutions reaches tipping point, which sadly it hasn't, then expect more fold-ups, broken dreams and 'big' boys grinning like crocs waiting for prey.

Best.
Alain Yap
Morph Labs

[Voice_Of_Logic]

I wouldnt trust my worst enemy's data to "the cloud". I hope this never, ever takes off. Not ONE client of mine will even entertain it and I'll tell you this - it wont be long before we have laws that regulate where my data (I own my information , thank you) goes. I have the final say. Its my info that these companies are throwing around irresponsibly and Coghead is just one more in the wheel of stupidy out there.

[inverse137]

Voice_of_logic, you are the blacksmith in the transitional age to the auto-mobile.

In the late 80s and early 90s the Internet was fragmented. We used search tools called Archie and Veronica. Then slowly the Internet evolved and became seamless. Now we have advanced tools like Google.

I have had 1 client ask about cloud computing and I told him it wasn't ready for prime time. But, I can see the writing on the wall. The "cloud," or whatever the final iteration is called, will be the norm. And probably sooner than we think.

While you may not store your data completely within the "cloud," your local network will be part of it. The "cloud" will jsut be another subset of the Internet like the WWW.

Data available anywhere, anytime is the future. Whle I am reluctant to move to a cloud concept I am not naive enough to think it is just a fad.

[Len Bullard]

When it was time to replace the horse and wagons and autos cost too much and there were no good roads, we built bus lines and trains. When it was time to make interstate travel realizable for supporting mass movement of troops and supplies, we built the interstate system, and then the car manufacturers bought up the bus lines and trains and tore them down. Now we need them again.

Cloud computing fits into a historical trend that trades control for affordable inconvenience until convenience becomes affordable.

[cgarrett]

I would agree if it didn't actually cost more than managing your on infrastructure at some scales. Otherwise, I do see the parallel between the cyclic thin-client vs. fat client debate.

[uusirna]

The cloud is the future. Bandwidth? It's growing and will continue to grow, driven by online video services. Server-side performance? No dedicated hosting can beat scalability of true cloud services, like AWS and Google AppEngine. Data security? It is coming - there are ways to keep server-side data secure (through encryption), and there are companies that start doing that.

[rapier1]

I love how people think that bandwidth is some sort of infinite commodity. No really, I do love it. It makes sure that the kind of work I do will always be in demand.