据Joshua McKenty，他曾在NASA的Nebula云工程变成OpenStack的基础之前领导过该工程，说云是计算机领域的下一个长期性革命。

第一次革命是大型机，然后是电脑，接着是互联网，而现在是云。

这是一个重大的宣言，也给了许多公司足够的理由，从思杰公司(CTXS)到Rackspace (RAX)，到VMWare (VMW)，再到红帽(RHT)，推出“云操作系统”并成为这个热点，新型计算平台中的“微软”。

据McKenty所说，问题在于，云不是一台电脑。它甚至不是互联网。它要复杂得多，更像20世纪60年代中期几乎使得IBM走向灭亡的大型机技术。

360系统问题教会了人们一些东西，Frederick Brooks，他参与了该项目，后来基于这些撰写了经典著作《人月神话》，也就是大问题不会对蛮力作出回应。让一个工程师加入到企业项目实际上会减慢进度，因为每个新员工在提供生产力之前都必须进行培训和监督。

McKenty说，该解决方案就是产生了Linux的开源过程。它耗费了Linux数十年时间来获取数据中心的主导地位，由于许多不同的团队从许多不同方向各个击破问题，每个团队的贡献最终构成了一个更大的整体，所以它起作用了。那也是为什么McKenty基于pentOS，从他在OpenStack这个开源项目上的Piston云启动并且没有要求它要是云问题的解决方案。

阻碍云的问题是任何互联网用户都很熟悉的问题。提供ID，获得认证，政策管理。使得这些事情变得无关紧要，自动化，天衣无缝是让企业构建他们会把自己堵上的云的先决条件。Nebula被转移到OpenStack上就是在寻找这些问题的答案，因为政府就是那种会完全利用完整云产品的企业。

对于投资者来说这意味着什么呢？这意味着，在短期内，诸如亚马逊(AMZN),谷歌(GOOG),和Rackspace此类的公有云公司就是可以押注的地方。基于成本来押注—成本效益最高的云会胜出。

这也意味着诸如Oracle (ORCL), SAP (SAP)和IBM (IBM)这样的企业软件公司有时间进行调整，惠普 (HPQ)和戴尔(DELL)也是如此。在这里投注，基于对开源工程的承诺，他说。

McKenty的结论是，尽管目前所有的那些关于云的炒作，这并不是1994年，那年Web破茧而出。这更像1989年，那时UUNet之类的公司定义了连接，CompuServe定义了服务，并且美国在线之类的东西被认为是最先进的。没有什么真正在互联网革命中存活下来了，而且预先对其中任何东西投注你都会一无所获。

McKenty认为当解决方案可用时，比起任何专有的筒仓（或半专有供应商）他们更像来自于OpenStack的开源过程，那时真正的暴雨才开始。

现在已经不是以前了。

According to Joshua McKenty, who helped lead NASA's Nebula cloud project before it became the basis of OpenStack, cloud is the next secular change in computing.

First came mainframes, then PCs, then Internet, and now cloud.

That's a big claim, and reason enough for lots of companies, from Citrix (CTXS) to Rackspace (RAX) to VMWare (VMW) to Red Hat (RHT) , to launch a "cloud OS" and become the Microsoft of this hot new computing platform.

Trouble is, says McKenty, a cloud isn't a PC. It isn't even the Internet. It's something much more complex, more like the mainframe technology that nearly destroyed IBM in the mid-1960s.

What the System 360 problem taught people like Frederick Brooks, who worked on the project and later wrote his classic The Mythical Man-Month based upon it, is that big problems don't respond to brute force. Adding engineers to a corporate project can actually slow progress, as each new employee must be trained and supervised before delivering productivity.

The solution, says McKenty, is the open source process that produced Linux. It took Linux decades to achieve data center dominance, and it worked because many different teams came at problems from many different directions, each contributing a little to the larger whole. That's why McKenty based pentOS, from his Piston Cloud start-up, on OpenStack, an open source project, and makes no claims of it being "the" solution to the cloud problem.

The problems holding back cloud are some any Internet user is familiar with. Proving your identity.Gaining authentication.Policy management. Making these things painless, automated and seamless are preconditions to major enterprises building clouds they will bet their companies on. It was in search of answers to those problems that Nebula was moved to OpenStack, because governments are enterprises that can take full advantage of a complete cloud solution.

What does this mean for investors? It means that, in the near term, public cloud companies like Amazon (AMZN), Google (GOOG), and Rackspace are the place to play. Place your bets here based on costs - the most efficient cloud wins.

It also means that enterprise software companies like Oracle (ORCL), SAP (SAP) and IBM (IBM) have time to adjust, as do HP (HPQ) and Dell (DELL). Place your bets here based on a real commitment to the open source process, he says.

Despite all the current hype concerning cloud, McKenty concludes, this isn't 1994, the year the Web was spun. This is more like 1989, when companies like UUNet defined connectivity, when CompuServe defined services, and when things like America Online were considered leading-edge. None really survived the Internet revolution, and betting on any of them beforehand would have guaranteed you nothing.

When the solutions are available, and McKenty thinks they're more likely to come from the open source process of OpenStack than the silo of any proprietary (or semi-proprietary vendor) then the real cloudburst can start.

Not before.