Joe Weinman is a global keynote speaker and expert of cloud computing economics. Having previously worked for Telx, HP, and AT&T, Joe went on to write Cloudonomics in 2012 and Digital Disciplines in 2015, yet still finds time to be on the advisory board of several technology companies. We had the great honor to sit down with Joe for a discussion about leveraging cloud computing for a competitive edge, as well as how the cloud could impact the world in the future.
About Joe Weinman
ML: First, could you tell us a little bit about yourself?
JW: My name is Joe Weinman and I’ve been with Bell Labs and AT&T Corporate, HP and Telx over the past few decades. Lately, I’ve been doing a lot of keynotes around the world. In fact, I’m doing a day trip to China tomorrow.
ML: How would you introduce yourself?
JW: I would say I’ve been active in cloud economics and helped define the field before it was even formed. I was talking about hybrid clouds and why they were mathematically optimal back in 2007 and 2008. I predicted the idea of dynamic pricing for cloud resources and people argued it could never happen. Then, Amazon introduced spot instances and people stopped arguing with me.
Lately, I’ve moved away from cost and performance optimization and into how the cloud can help achieve strategic competitive advantage in digital technologies. That’s what I talk about a lot and that’s what the new book is about.
ML: To those unfamiliar with the cloud computer, can you tell us a bit more about the hybrid model?
JW: Sure. There are a lot of misconceptions about the benefits of the cloud. What makes things even more complicated is that these benefits vary based on who you are. For example, if you’re a small company then, chances are good that large providers will have a much better cost structure than you can. Therefore, you can’t imagine a pizzeria having a data center behind the pizza oven and staffing it full time with the world’s greatest IT talent. On the other hand, if you’re a large enterprise with financial resources and the ability to attract talent, then there’s a case to be made that you can achieve a similar and sometimes better cost structure compared to a large cloud provider.
I’ve looked in depth at a variety of drivers for costs. When you actually look into it more, it turns out that these drivers are not as obvious as they might seem. For example, one of the effects that happens is statistical multiplexing from uncorrelated workloads. I’ve done some characterization of penalty functions associated with diverse uncorrelated workloads, and it turns out that those benefits cut in at relatively small quantities of workloads. Even if you only have 2 workloads, you can achieve a 30 percent reduction in the penalty costs assisted when under or over capacity. Once you get to 100 workloads, you’re at 10 percent of that nominal penalty. That 10 percent is more than made up for by other kinds of advantages, at least as far as infrastructure costs go.
It’s a little bit counterintuitive, because people have argued about the enormous economies of scale and the cloud. The other thing that’s interesting is that there are performance advantages from running on custom infrastructure. There, you’re playing a tradeoff game around being able to leverage elastic standardized infrastructure and being able to exactly customize something to your needs. Tradeoff happens in every other industry as well.
There are all different kinds of advantages and disadvantages, but one of the biggest ones perhaps is that a cloud provider is a profit-seeking business. Consequently, you can’t just look at relative cost structure. You’ve got to also look at the difference between the price that a cloud provider charges versus the cost to do it yourself.
ML: We tend to think about the cloud as something needed for our infrastructure, but in your book, you also talk about competitive advantage strategies. Can you tell us how leveraging the cloud can become a competitive advantage for a company?
JW: There are many ways in which the cloud can be used not just for optimal cost and optimal architecture, but for business as well. The basic argument is that there are a number of ways in which the cloud can provide better customer value. Differential customer value, in turn, can create competitive advantage. As well, the cloud itself can offer competitive advantage without necessarily being directly linked to customer value.
The user experience can be critical to competitive advantage and value creation. Some studies show that by reducing page load time by a couple hundred milliseconds, they were able to increase revenue by 8 – 13 percent. There are several reasons. One is that if pages take too long to load, then customers aren’t sure if they should wait or if the site is down. Pages take too long to load, and the customers say, “You know what? This will be faster if I go to the competing site.” If you have a global user base and you’re trying to maximize the customer experience by minimizing response time, you want to have resources as close to those customers as possible. You can build your own facilities very close to the edge, but in a case like that, you now start getting into issues where you’re chopping workloads up into multiple locations. The best solution would be using elastic cloud resources.
About Digital Disciplines
ML: How do you describe the “cloud” from a general perspective in your book, Digital Disciplines?
JW: The cloud is the architecture required to implement many functions that have turned out to be critical to revenue and profitability. An example is customer reviews. In the old days, I would tell my neighbor I just bought this grill or blender and it really sucks, or it’s really great, and I would impact a handful of people. Now, with cloud-centric social media, whether it’s Twitter, reviews at Amazon or the manufacturer’s site, the cloud is the essence of providing recommendation. People rely on credible, unbiased recommendations from both random strangers and their social networks as key sources for decision making.
There’s another case where the cloud, generally speaking, is very relevant to profitability and revenue growth in competitiveness: the area of innovation. I’ve looked in depth at innovation in my new book, Digital Disciplines, and there are many ways in which the cloud and its approach tends to be critical. It’s everything from collaborating with a global team of designers and engineers and being able to co-create with customers, to leveraging the cloud for cloud-based contests, challenges, innovation networks and idea markets. Companies that leverage the cloud in various phases of product or service development can basically get to market more quickly. By leveraging functions, whether lambda functions or platform service components, they can more rapidly assemble the functionality and have it be higher quality. Thanks to the cloud, we get cheaper, faster, better innovation and product release cycles. That’s before we even get into movement, continuous delivery, and pivoting with the idea. The faster we can iterate means we can get feedback and change course faster. It’s the difference between a speedboat maneuvering through waters versus launching a cannonball and hoping that you aimed it properly.
ML: In the book, you talk about appropriate cloud use cases. Could you talk about cases when companies misuse the cloud?
JW: You wouldn’t do data compression in the cloud to minimize network costs, because you need to have the data on your own premises. You wouldn’t want to transport multiple petabytes to the cloud to compress because you can presumably compress it better and avoid the network. Obviously there are some tradeoffs.
Another thing you wouldn’t want to do is encryption in the cloud, in the sense that you wouldn’t send un-encrypted data in plain text over a network so that it could be encrypted in the cloud. I go through lots and lots of use cases and there are different mathematical principles at work. For example, rather than having multiple copies of data, it’s beneficial to have a single repository in the cloud or perhaps a single logical repository that may in fact have multiple replicas for business continuity and faster recovery purposes. That’s a great case where it makes sense to consolidate as long as response time is not an issue.
ML: I recently read your article, “The Internet of Things for Development Economies.” When consumers think about the Internet of Things, we think about smart thermostats, refrigerators, and efficient vehicles, but in your article you talk about how the Internet of Things can impact the developing world. Can you talk more about that?
JW: I think there are a lot of different opportunities in terms of all the basics we take for granted like food, drinking water, energy, and healthcare, where the Internet of Things will make a huge difference. I think it’s a great potential payout for society. It’s not just the usual profit and revenue kind of thing that business look at. It’s a matter of human welfare.
For example, imagine smart water pump handles. For a lot of the developing world, a child might have to walk for an hour to get to the nearest well and then walk back. That well may be dry, in which case it is a wasted effort. Some smart pump handles monitor water levels. Perhaps, over time we can route people optimally. In some cases, it literally may be a matter of life and death.
Another good case is the cold chain and refrigerated vaccines. You may or may not know whether a vaccine is still useful because, if at some point it’s not refrigerated for two days, it may become inert and useless. You’re still trying to inoculate people and wouldn’t necessarily know that it wasn’t useful. An interesting technology could be a little strip that changes color when the vaccine goes above a certain temperature for a certain amount of time. But, you couldn’t fix the supply chain process and respond in real time if you didn’t know that there was always a particular error code in a particular location where this problem was happening. Let’s say a refrigerator had been working fine but there was a power cut and you realized it would make more sense to have a plane carrying those supplies to a different distribution center to ensure the vaccine survived delivery.
ML: There’s Google, Big Query, and other engines that allow us to get the value of analytics. How do you think the impact of big data and the power of big data will continue to impact and grow over the next few years?
JW: I think there are several basic concepts or trends at work here, one of which is “more data is always better.” As the cost of data capture and data storage continue to drop exponentially, it enables all sorts of things that people couldn’t think about even a few years ago.
In my worldview, big data is applicable in a broad variety. One good example is its relevance to process optimization, which can generate customer value. It can also generate firm value, which is obviously farther removed than “people who bought this also bought this.” That’s the first thing that is relevant not just for what I call “collective intimacy” and “customer intimacy,” but also for better processing and resource use. As well, you can build better products and services that have more data, which these products can utilize to improve their functionality in a variety of ways.
For example, John Deere grain combines enable farmers to optimally traverse a field which may be irregularly shaped. The trick is you want to minimize fuel consumption and time while exactly developing the optimal path through a field. You don’t want to double harvest the same square foot, and you don’t want to miss anything that you can sell. It’s a case of being able to leverage data which can include weather, satellite imagery on the state of the field, or maybe infrared. It’s a whole set of external data that can include GPS, weather and so on, but also data which may be internal, like the scheduling of farm workers. Wherever I look, it seems like things can benefit most from a mix of internal and external data.
An example of this would be doing a market-basket type of analytic. It helps you to know not just that people bought more beer on Tuesday, but that they bought more beer because that day was a big playoff game and the weather wasn’t snowy, so they decided to have parties. That’s a case where massive amounts of internal data are great but also having data on context is important.
ML: One last question. What’s on the horizon for you over the next few years? What exciting projects are you working on? Can we expect another book?
JW: Right now, I’m doing a lot of speaking, but have started to think about doing a book on disruptive innovation. I think there are some issues with the current model of disruptive innovation, and I think I figured out the correct model. Obviously, that’s a little bit beyond the cloud and technology, although it can be very much tied in with the ways the cloud can help with disruption. If this book ever comes to fruition, you wouldn’t necessarily know that I had any background doing stuff in the cloud. That’s exciting.
Resources
You can visit Joe’s personal website and check out his books:
