Most Read This Week
Enterprise Cloud Computing
What’s Missing from “Cloud First” in the Enterprise?
Cloud First doesn’t just apply to applications, it needs to apply to infrastructure as well
By: Brian McCallion
Mar. 9, 2014 04:00 PM
My first experience with an "inverted yield curve" was in 2000 just prior to the tech bubble bursting. I was working on a financial portal for an investment bank and one of the charts was a yield curve. It looked odd all of a sudden, so I looked it up in a book of financial terms. An inverted yield is indicated when interest rates for short-term capital are higher than interest rates for long-term capital. In other words, people are willing to pay a significant price to alleviate short-term concerns because they're focused on the now and not so concerned about one year, three years, five years, or thirty years from now. Inverted yield curves some believe signal disruption in financial markets. On the surface, Cloud First seems to signal the disruption that is cloud computing. To take this metaphor a little further, this inversion of Cloud First from "Cloud Never" suggests to me an inverted set of concerns. Does Cloud First prioritize an immediate need to say "something" about the cloud and cloud strategy? Does Cloud First prioritize the now while discounting near, mid, and long-term opportunities that far exceed the "costs less, more agile, faster time-to-market" recording I hear played daily throughout the blogosphere? Beyond saying "Cloud First!" what else can enterprise technology teams prioritize that may amplify their ability to execute in the cloud?
One of the primary strengths of "Cloud First" is the simplicity:
Q: "Hey, where do we deploy the new financial system?"
A: "Put it in the cloud."
Yet the real power of Cloud First lies in the underlying recognition that details and choices, as they filter down through multiple levels of management, quickly become confusing. On many levels in corporate IT I get the sense that the more details provided, the more likely things will go forth along an unintended arc.
Cloud First reminds me of the trailer from the movie "Face Off" starring Nicholas Cage and John Travolta, in which they actually do trade faces for a while. The best thing about the title of the movie is that you know something pretty amazing (and possibly horrifying) is going to happen and it grabs your attention. Cloud First? - same thing - it's got everyone's attention: we're unleashing the monster the media has been warning you about and it's too late to turn back. If we need to do a sequel we'll call it "Mobile First."
What Does Cloud First Look Like?
What Does Cloud First Lack?
Whatever cloud you're on, try to consume cloud services. As an architect, specifying services rather than asking teams brand new to cloud to quickly build robust, highly available databases that will withstand the rolling outages of a year like 2012 is really unfair, and it just won't happen. It's like the scene in "Kill Bill" where the "Crazy 88" suddenly demand 20 pizzas in a Sushi restaurant. It will just exasperate people and make them go, well, crazy. In cases where an Oracle RAC database was absolutely necessary I solved the problem by defining an architecture in which the application layer ran in the cloud, but the database ran "close" to the cloud via a low latency fiber cross-connect. I don't suggest you try this unless there's no other option, as was the case in 2011 when I worked on that solution.
In other words, it's really difficult for IT teams in large corporations to suddenly build highly available databases in the cloud. I've seen it end very badly even when implemented by good people. In many cases the services available in the cloud have the scalability and availability "wrapped" into the service. Similar to the way Linux succeeded largely because "with enough eyes all bugs are shallow," perhaps the cloud manifesto is that given enough implementations and users, a cloud services performance and availability will blow away a one-off, bespoke database implementation built by a team new to cloud. Not building services from scratch may not be as much fun or as hard core as what the cool startups do, but you'll have plenty of interesting things to figure out without building everything from the ground up. One of the rules of Cloud First is to focus your team's energy and avoid fighting battles on soggy unfamiliar ground. One of the more subtle messages of Cloud First is that a key element of a corporate cloud strategy is to avoid building stuff from scratch unless you have zero other options. People may work very hard to convince you to do otherwise, but stick to cloud services first.
Cloud First Doesn't Just Apply to Applications. It Needs to Apply to Infrastructure as Well
Reader Feedback: Page 1 of 1
Subscribe to the World's Most Powerful Newsletters
Today's Top Reads