Cloud computing's killer applications
While 2010 was clearly the year in which cloud computing defined itself, 2011 is shaping up to be the year in which cloud clearly demonstrates what it is good for.
The year 2010 will probably be remembered at the year that cloud computing "shaped" itself into a tangible concept, at least amongst those of us who care. 2011, on the other hand, will likely be the year in which IT figures out how to actually use cloud concepts.
Of course there are success stories dating back two or more years, but what is happening so far in 2011 is a growing body of businesses, data, and applications that were born and cultivated in the cloud. Add to that the online and conference communities forming around cloud and new application categories and we are starting to get a clear picture of what the "killer applications" of cloud really are--at least as of today.
I want to step through the three that I think are the most impactful application categories for cloud and show some examples of why cloud enables these applications to exist:
Data collection and analytics
There is no doubt that one of the biggest revolutions enabled by the cloud is the ability to store and--perhaps more importantly--process large sets of data, either as very large scale fixed-length jobs or in real time. The economics are simple: without the large upfront capital expense, the total cost of using a large number of systems for the period of time it takes to process the data is significantly reduced; and "downtime" costs nothing more than the cost of storing the data.
I attended O'Reilly Media's Strata conference in Santa Clara, Calif., last week, which covered "big data" and the various applications of that concept to a variety of problems. I walked away extremely impressed. The innovations on display were mind-blowing. It really felt like a Web conference circa 1999, with talk of almost endless entrepreneural opportunities for those with the right ideas.
I attended sessions on data journalism (which demonstrated how data analysis can be used to find key facts regarding the news of the day) and the application of data capture and analysis to government and academia. Both sessions demonstrated applications that amazed me, including several that used public data sets to give new insights into how we think and act as a society.
Personal data analysis was also on display, such as Strata start-up showcase winner Billguard, a service that will analyze your credit card bills for potentially abnormal activity by combining your statement data with other data sources and "wisdom of crowds" tools. Services like Billguard lead me to believe that consumers will benefit most from "big data" innovations.
The cloud has clearly changed the game for data and it looks like data may in turn may a central role in the future of cloud computing.
Online commerce and communities
This is probably the category that most followers of cloud computing would think of first when they think of killer applications for cloud computing. There is no doubt that cloud has changed the game for Web applications and services. Online services and communities are appearing at a dizzying pace today in large part due to the economics of cloud computing.
How is that, you might ask? Well, again, its because the cost of failure is so low in cloud. Want to try a new business idea, or play with a new online service concept? You have plenty of options for building, hosting, and scaling your concept, all of which are on a "pay-per-use" fee schedule--which ultimate means that if you fail, you stop paying and don't have sunk capital costs to offload or absorb.
This is why the Silicon Valley--and likely worldwide--venture capital community has so quickly changed the nature of financing for online start-ups. It is now almost impossible to get a VC firm to buy you a server. Rather, they'll tell you to use cloud services to build and test your business concept.
If your idea shows some signs of success, they'll then tell you to scale the business in the cloud--again, to avoid sunk capital costs. If at some point your business is wildly successful, and a private data center or fixed hosting agreement starts to make more sense, then your VC board members will probably be happy to tell you to take some of your own cash flow and do what you need to do.
The point is that services such as Google App Engine, Rackspace Cloud Files and Cloud Servers, Amazon Web Services, and a laundry list of others have made delivering new software and business concepts to the online market much, much easier, and the associated risks much, much cheaper.
Context vs. core
What may surprise many is the increasing adoption of cloud and hosted options for "context" systems (in the "core vs. context" paradigm introduced by Geoffrey Moore in his management books). In short, context systems are those that may or may not be mission critical, but add no real value to the distinct business that they support. Some examples of context systems include e-mail, telephony, and document management.
In my role at Cisco, I've seen an increasing number of customers declare goals to freeze or reduce the number of wholly-owned data centers that support their businesses, choosing instead to find online services to meet much of their needs. Where they all seem to be starting is those context systems, with vendors like Salesforce.com, Microsoft BPOS, and several smaller SaaS vendors benefiting.
So-called "core" functions--which are those functions that a business' competitors would find difficult to replicate--are taking much longer to move to the cloud. This isn't surprising, considering the sensitive nature of the data and code associated with these applications, as well as the investment made into existing systems by enterprises.
While I believe core systems will move to cloud models in the coming years, I suspect that legal, financial, and technical issues will make that a much slower transition than context applications. That said, familiarity with cloud through those context apps will probably accelerate the rate at which enterprises become comfortable with cloud models, so who knows.
The other "context" function greatly benefiting from the move to cloud models is storage. Backups, disaster recovery, object storage, content management, and many more storage-centric functions are moving to the cloud at a rapid rate. While security remains a consideration, the introduction of encryption and strong authentication technologies are rapidly expanding the market for storage services.
While I have no doubt that other forms of enterprise computing will move to the cloud over the coming year, these are the three categories that stand out for me. What about you? What do you see as the "killer apps" for cloud computing models?