Brian Katz thinks the cloud is “a pile of crap.” He laments the incessant debate about cloud or not cloud, private vs. public, and so on. Some days I’m inclined to agree with him. There’s certainly plenty of geekturbation going on. I have disliked the word “cloud” since the very first time I heard it. I have been an unrepentant advocate for the alternative term “utility computing.” Cloud describes an implementation detail. Utility computing explains what it does and why it’s useful. Features and benefits: marketing 101.
Brian wants us to focus on customer value. I complete agree with that sentiment. I don’t agree, however, that cloud is of no interest to anyone outside of IT. On the one hand, the fact that Netflix runs their infrastructure on Amazon’s AWS platform is their own concern. On the other hand, though, the fact that Netflix can provide consistent performance in the face of massive usage spikes and small, medium, and large system failures, is entirely relevant to their customers. Users may not care about operational implementations, but they do care about operational excellence. Furthermore, if Netflix tried to implement the same level of operational excellence with data centers full of idle gear, their streaming service would cost a lot more than $8/month.
Netflix goes even further. They make deep technical decisions about things like threading and API-call failure handling from the perspective of the impact on the customer. User-centered IT doesn’t get any better than that. The fact that we waste our days arguing on Twitter doesn’t obviate the intimate relationship between technical architecture and customer experience.
Brian and myself are among those calling for IT to be more responsive. Cloud computing enables IT responsiveness. One of the current cloud debates concerns Amazon’s AWS offering, its openness or lack thereof, and its exact relationship with its competitors. Amazon is succeeding because they are relentlessly focused on providing value to their customers: SaaS developers. It seemed like only days after I’d helped a client tip up a memcached server on AWS that Amazon introduced their ElastiCache service. I felt like they’d read my mind. If Amazon does achieve lock-in, it will because they read their customers’ minds better than anyone else.
Even if we posit that Amazon is the loveliest of cloud providers, Brian’s question remains: who cares other than IT? On one level, no one. On another level, though, AWS helps SaaS developers deliver value to customers faster, with greater quality and less investment. Service design researchers Robert Glushko and Lindsay Tabas point out the need to integrate front and back stages in order to create good customer service experiences. For the most part, the cloud is back-stage. I don’t think, though, that you can divorce it from front-stage in terms of the overall value proposition.
SaaS, on the other hand, is front-stage. Lori MacVittie and others don’t consider SaaS to be the cloud. I partly agree with her. Many software services are just Web 2.0 apps. But let’s take the example of Mozy. What’s the difference between Mozy and Time Machine? Both of them do exactly the same thing: “back my stuff up every night at midnight.” With Time Machine, you have to buy and provision backup storage. With Mozy, you don’t. With Time Machine, if you only need 10 GB of space, and the smallest version of the Time Machine server is 500 GB, you pay for unused capacity. With Mozy, you pay for what you use. Mozy brings utility computing (consume based on demand, pay based on consumption, someone else manages) to the small business and the consumer.
I find “lean” to be a helpful way to think about cloud computing. Lean puts the focus on increasing value, decreasing waste, and minimizing Non-Value-Add (NVA) activities. Providing customers with consistent application performance in the face of spikes and outages increases value (remember, customers care about operational excellence). Not paying for idle servers decreases waste. Letting someone else rack servers and swap hard drives minimizes NVA. At the PaaS level, you can say the same thing about not having to configure load balancers. At the SaaS level, you can say it about not having to maintain on-premise backup hardware. It occurs to me that “Lean Infrastructure” may be a proper term to apply to cloud computing at all levels.
In sum, Brian is probably right: we probably spend too much time debating arcana. At heart, though, I think we’re at least trying to argue about the right thing: how to enable organizations to provide the most value with the least waste. At the end of the day, I don’t think it’s enough just to call it “computing.” Instead, I think it’s about trying to achieve “lean computing.”