The transition from “Old IT” to “New IT” is accelerating. And, the discussions about what this means to organizations can be passionate at times ̶ particularly around cloud computing. We are increasingly hearing terms and concepts being tossed around about the role of IT as it relates to cloud computing – terms such as service integration, commodification, and loosely-coupled.
One of the most prolific speakers on cloud computing is Peter Coffee, vice president for strategic research at Salesforce.com. Peter has been a frequent visitor to Canada over the past several years speaking at cloud conferences and listening to him is always enlightening ̶ and surely entertaining. Forward-thinking businesses are adopting innovative cloud computing platforms like the Salesforce and Box to gain a competitive advantage. Think deeply and amalgamate cloud computing’s technology disruption in the context of your own current business initiatives and decision making.
Brian Clendenin: What do you think of the ‘New IT’ vs ‘Old IT’ phraseology being used today?
Peter Coffee: “I think an important distinction between ‘New IT’ and ‘Old IT’ is that ‘Old IT’ began with a conversation about the capital budget cycle and you couldn’t even think about doing something until you acquired and deployed the capacity and that meant a minimum of one to three or four quarters before you could get serious about starting to solve the problem. Now, some people believe that a so-called private cloud model will obviate that because now you just spin up some virtual servers. But, you still need to own the capacity… and that means that someone has the incremental project that pushes you to need a whole new rack of servers or maybe even a whole new room of data center or maybe even a whole new building because you’ve reached the limits. Many corporations today are discovering that the expansion of a physical data center is bumping up against limitations of power supply, cooling load and so on. You can’t just take a spare conference room and put in a raised floor and put in some servers anymore.”
Clendenin: What comes to mind when you hear ‘we want to do it ourselves?’
Coffee: “When people say, ‘but, we want to do it ourselves,’ the fact is, doing it yourself is capital intensive, skills intensive, and innovation poor. And I don’t think anyone has that much capital, that much talent or that much willingness to let the competition out innovate them that they should be happy about that idea.”
Clendenin: Another term being used within the cloud computing world is the notion of commodification. People have differing opinions of what it means. Clayton Christensen, author of The Innovator’s Dilemma uses the steel industry as an example of what it suggests to innovation. What does commodification mean to you?
Coffee: “Commodification is used by a lot of people to suggest that something’s going to wind up being bad. Well, you know, steel beams are a commodity because they’re produced to a specification and are completely substitutable for each other. It doesn’t mean it’s bad steel, but it means it’s completely substitutable. And, the National Institute of Standards and Technology has propounded an idea of cloud standards that’s a lot like a steel spec. It’s the idea that, essentially, a running workload should be able to move from Fujitsu space to IBM space to Hewlett-Packard space to Amazon space and never even know that its moved. Well, we’ve seen that model in another era and it was called the Wintel model of desktop computing where there was very little room for innovation. I had an Intel vice president at the time the 486 was introduced say to me (when I was a reporter at eWeek) that engineers would get to pick the color of the box … that when you looked at the level of integration at the system on a chip aspect of these advanced Intel chips and with Microsoft insisting that it owned the visual copyright to the Windows desktop there wasn’t really a lot of room for innovation. And, really, we didn’t get much innovation in the desktop computing space during that period of time. That’s what commodification means to me.”
Clendenin: Let’ switch gears and talk about the frequency of updates that is part and parcel with cloud computing.
Coffee: “Sure … the new world of updates. No one apologizes for having a rapid cycle of updates. If your quality is low people will complain, but having frequent updates is now almost a figure of merit for an application. Where in the old world I have to install, regression test, re-implement customizations and so on. The typical Windows user dreads the prospect of making a complete version change. They know they’ll be re-installing applications, they know they’ll be de-bugging option settings that didn’t make the transfer properly.”
Clendenin: That sounds like an onerous and time consuming process.
Coffee: “It’s a fraught moment and, at the enterprise level, it’s even more so to upgrade a major piece of enterprise software like an ERP or a financials. It’s something that you wait for years to find the chance to do it. And, I know major global brands that do these things over Memorial Day weekend or something because that’s when they feel like they’ll get the breathing space to get it done. What a horrible way to think innovation compared to our model (cloud) where we follow the sun ̶ one 24 hour cycle. Essentially, 100% of our customers choose to accept an upgrade. And, they rely on us to have massively, exhaustively tested to be sure we won’t break anything. And, we’re able to do that because we can’t see their data, but we can still exercise their code and confirm that there’s been no operational impact that would break anything. That’s really quite a unique change in the way enterprises think about their IT and about the process of improving it.”
Clendenin: From an IT strategy perspective, it’s important to understand that new application design is modular and API-based along with being ‘loosely coupled.’ How would you explain loosely coupled to someone unfamiliar with the term?
Coffee: “Loosely coupled is an expression that’s used by software developers. Loose coupling implies that it’s now much more feasible to change a piece of a system and to know exactly what you need to preserve not break anything else. In the old world of software where you didn’t have good object level of abstraction, any change you made could conceivably result in something that was looking at a particular memory address no longer finding what it expected to see there. And, this is a fundamental problem in the old world of the PC where programs were essentially written in machine code for acceptable performance and acceptable memory use and the result was that something as trivial as moving to a different display would change the memory map of the display buffer and all of the sudden things just wouldn’t look right – that’s very tight coupling. Loose coupling is at the API level and at the ‘pointers-to-pointers’ level and its where a lot of the performance, the raw performance at the CPU level has gone over the last two years, but the result is that now you can do things like just tap the ‘Update All’ button on your iPad and then drink your cup of coffee and really not worry that your apps aren’t going to break – which I find, personally, a big quality of life improvement.”
How much longer will we have to keep inculcating the business benefits of cloud computing to enable organizations to be more productive and competitive? I don’t know the answer. Organizations are at different levels of maturity in their journey to the cloud. One thing is clear … Major technology shifts, like the one we are experiencing now, only happen every 10-15 years.
Embrace the exciting times we are living in and focus on the business value of IT.