The cloud has a silver lining after all
Connecting state and local government leaders
Once you understand what it is and is not, the move to cloud computing represents an opportunity for a disciplined approach to tackling information overload.
At a recent executive breakfast panel, I was asked which information management strategy is right in this rapidly changing IT landscape. With the cloud computing cheerleaders out in full force — such as in Microsoft’s goofy “To the cloud!” commercials, yet more evidence of the “cloud is everything” mentality as discussed in my last cloud computing article — this is both a very common and relevant question. Let’s try to answer that question with special focus on the ramifications for federal agencies that are moving to the cloud. There, you see, I said it: “to the cloud!”
Let me start by saying I see the transition to cloud computing as a good thing despite my previous articles on the immaturity of cloud standards. Over time, the standardization issues will be resolved, in either a formal or de facto manner. However, that issue is orthogonal to the challenges of transitioning existing applications to the cloud.
Related articles:
It’s time to burst the cloud’s hype bubble
Why cloud computing is still a red herring
To understand that process and the critical decision points, you must understand, at a technical level, what the cloud is and what its origins are. For example, the cloud is not equivalent to the Internet — as I saw asserted in a reader’s comment to one of my previous cloud computing columns. Cloud computing is a distributed computing model for applications; the Internet is a communications platform. Just because cloud implementations can be accessed via the Internet and possibly share the characteristic of seemingly unlimited elasticity, it does not mean they are one and the same. Similarity is not equality.
And the cloud’s origins? Although there were multiple converging influences that led to the birth of cloud computing, such as application service providers and Salesforce.com, I would contend that its true founding influence was Google’s published papers on how the company’s index of Web pages scales to thousands of computers.
One paper discusses the parallel processing algorithm called Map Reduce, and the other paper discusses the implementation of the Google File System, which allows files to grow to arbitrary sizes across any number of computers. These papers were quickly followed by open-source software, such as Hadoop, and commercial implementations, such as Amazon Web Services.
The influence and importance of those new concepts cannot be overemphasized because they represent a whole new shift in computing: a multimachine-based operating system. Although distributed computing has been around for a while, cloud goes way beyond that because distributed computing operated at a low level in which network communication was required.
Cloud computing abstracts distributed computing to the next level, into a cohesive, single virtual computer that spans any number of physical computers. For example, the Google File System and the Hadoop File System enable arbitrarily large files to be stored and processed across thousands of computers, as if — from the programmer’s perspective — they were on a single computer. That is where cloud computing is taking us: the elasticity of thousands of computers that can be programmed just as easily as programming a single computer.
With an understanding of cloud computing beyond just “to the cloud,” we can now explore the real issues around migrating to the cloud.
Simply stated, migrating applications to the cloud, also known as platform as a service, or PaaS, requires re-engineering them from scratch. Why? Moving from a single-machine paradigm to an elastic-machine paradigm requires a rewrite. To be clear, I am only considering PaaS here, and the ramifications are different for the other flavors of cloud computing, such as software or infrastructure as a service.
NEXT STORY: On the hunt for great public-sector websites