I used to get this question a lot at my current old company.
” How can we, as a user of several operational systems, take more advantage of The Cloud?”
This is where the infamous Gartner Hype Cycle, of which there are hundreds of examples, should be referenced. In terms of The Cloud and its reference to this chart, it’s getting higher and higher to the ‘Peak of Inflated Expectations’. I agree with this for the most part. As more technology companies commandeer the term for marketing purposes or shove some old/new software package onto EC2, expectations get set. Commonly the setting is wrong. Marketing philosophy aside, I find myself explaining what the cloud can actually do, what its actually used for to more people I’d (shall I say) expect better from.
I recommend starting here for a general semantics overview to describe all the aspects of the cloud. There is no One Cloud for all things, all people. I agree with the 3 layers of cloud services comprising of SaaS, PaaS, and IaaS. The mainstream cloud service of Amazon Web Services is an IaaS and PaaS model, depending on which service you buy.
I think for corporate entities that have applications that are not web properties or scientific/analytic computations, the options for leveraging the cloud are relatively slim. Stated more directly, the value proposition of the cloud is minimal for most applications. Anyone can put anything “in the cloud” but if the application isn’t built for it, you are simply leveraging a pay-as-you go managed service.
I have only seen 2 basic services that my old company could have leveraged for their “operational systems”. Database Backup service similar to Zmanda (that Ingres would of course build their own – I would expect) and Outlook attachment offloading. I am not including already purchased SaaS solutions such as SF.com or Intacct. Thus, these little niche solutions are not necessarily game changing use cases for the cloud.
Another alternative is to simply put demo environments on the cloud for field reps, allowing them to spin up instances when they need it without crushing their laptops with VMware or something similar. Finally, you could argue QA or performance tests of some type could be done, although the performance tests would be narrow in scope and probably cause most performance engineers to puke at the idea. Again, sort of a ho hum advantage.
This is where the cloud and its messaging will need to go through a marketing maturity model of sorts. The current drum beat that the Cloud could save energy issues, eliminate system admins, feed a starving child in Bangladesh, etc. etc. all need to be set in context to real business value.
A recent webinar I attended hosted by Information Week took a poll of the attendees asking their personal interests in cloud services. 50% said they were “checking it out, still waiting to jump in”. For all the hype and attention, most see it as new fangled technology worth a peek, but thats about it. This will continue, in my opinion, until applications are (re)written to take advantage of the cloud infrastructure seamlessly(like Mathematica) or ground up cloud apps which most Web properties seem to have already conquered.
I attended a conference last spring(2008) where Simon Crosby(CTO of Citrix / Xen) spoke about the future of Hypervisors and why everyone else was doing it wrong. Entertaining talk, although I thought he was going to literally pull his hairline back 2 inches with all the double handed hair pulls. Virtualization is to clouds like water is to farming, its a must have. The piece that stuck with me was the current lack of instrumentation and toolkits for Hypervisors to understand the resource load requests of applications *before they request it*. He went on to say that in order for more packaged apps to work with VMs and Hypervisors intelligently, more sophistication(and a standard) are needed to collect the needed meta metrics for a true dynamic elastic response of resources such as memory, CPU, disk, etc. . Some applications do this today, but there is no standard, no toolkit for developers to write applications to respond to the benefits of the cloud. Just like a plug in the wall, the electric load ebbs and flows depending on what I do. All the “apps” writing to those plugs fit a standard API and thus the grid responds predictively.
The talk was ages ago, so I do wonder what new is happening in the application development space to enable ISVs and customers to write “cloud aware” applications?
I think time and start up investment will change this in 2010.