Recently I had the pleasure to sit down with Darren Person, VP Chief Architect at Elsevier. We discussed his perspectives on the cloud computing industry at large, what developments he is interested in, his perspectives on leading an IT organization, and how Elsevier is approaching the cloud as an enterprise.
Gathering Clouds: From a very high level perspective, how are you thinking about the cloud in the context of your position at Elsevier?
Darren Person: We think about the cloud as the natural evolution of traditional computing through integration and automation. Some of the most important drivers of the cloud are speed, agility and cost. On an ongoing basis, there are lots of new ideas, new products, and, inherently in a large-scale enterprise, a large amount of legacy. When we make the business cases for our new products, typically that comes with upfront capital costs, slow lag and long delivery times without a solid understanding of the potential outcomes (e.g. will this new venture be successful?). The cloud gives us the ability to “fail fast, fail cheap” or “succeed fast, scale quickly.” In addition to speed and agility, we see tremendous value in the new operational paradigms the cloud enables, what we and the industry have coined as infrastructure-as-code: the idea that our applications can be self-monitoring, self-managing and self-aware, while dynamically scaling or self-healing, depending on the attributes in play. In our traditional models, we couldn’t do that because we didn’t have excess capacity just lying around nor could we dynamically allocate and deallocate infrastructure on-demand. In IT, cost management is a critical aspect of service delivery and the cloud makes the ongoing operations cheaper as well as reducing upfront capital outlay. As we’ve been doing increased cost analysis in the cloud, it’s clear that there are substantial cost savings opportunities in the way we architect our applications, in the way we think about our applications, and the way we deliver those applications to our customers.
GC: How do you reconcile the disparities between the needs of your internal IT organization with the requirements of an outsourced cloud infrastructure? What does that mean in the way you structure your teams and how does that impact the way you look at integrating outside vendors into your environment?
DP: For a long time I’ve felt that the vertical silos of an IT organization are inefficient and typically are “IT begetting IT”. So, what do I mean by this? If you look at IT departments historically, when you separate IT into 2 separate verticals, infrastructure and development, you are pitting 2 organizations with a shared goal and responsibility against one another. So the common characteristics of these organizations include non-productive arguments and large-scale debates about accountability– with the ultimate answer being, everyone is accountable. What we end up with is a lack of innovation and an overabundance of internal distractions keeping us from achieving our goals.
When I look at the cloud, I see it as a convergence opportunity. To me it isn’t only a platform to enable agility, quicker time-to-market and cost savings, but also an opportunity to do convergence where the infrastructure teams and the application developers are level set into a single world where everything is built through API’s and a common language, “code”.
GC: Do you view cloud as central to this evolution of IT, and moving forward, what do you see IT’s purpose being in the future, with regards to how it integrates cloud, what it means for the organization, and so forth?
DP: There are many coined terms in the market now that weren’t there just a few short years ago. But the DevOps movement, from an organizational perspective, is a highly attractive way of running things, converging the responsibilities of your engineering team, both on the software and administrative sides, into a shared responsibility for production. With shared responsibility, your outcomes are typically higher reliability, increased availability, and improved performance for your applications because you have everyone thinking of these components as part of the architecture of the development processes. What I see now is that we are starting to remove a lot of maintenance tasks, where IT spends a large percentage of its time focusing in the “keep the engines running” mode. As we start to move those tasks off to an outsourced partner or to the cloud, I think this provides a new opportunity for technology to step up in the organization and start taking a lead role. The reality of the situation is clear: technology each year is more pervasive and is leading the industry more than some of the traditional business models. A lot of what technology enables is disruptive across many markets. So the more tech people, those who understand the technology and its use are involved in driving the business and innovation, the better positioned those organizations will be to come up with those competitively differentiating services. As I look ahead, I see technology people spending more of their time supporting the business – conceptualizing new ideas and business models, and applying their tech experience – versus where they spend their time today.
GC: Elsevier is a huge company, a well-known brand and an innovative company. But you are bringing internally a paradigm for IT which presents a breakdown in terms of cloud adoption at that scale of a business level at the enterprise or SME level, what is your perspective on why it is working for you guys? Is it because you are an evangelist of cloud or you see its value in the context of your broader organization? Or, is there something about Elsevier’s model that makes it more cloud-ready than some other larger businesses?
DP: Its two things.
One is that it’s a natural evolution of a data services business. While Elsevier is a publishing company at it’s core, a lot of the products and services we offer are online. We leverage our content, apply cutting edge technology and serve it to customers to add additional value to their resarch. I think at the onset of that, the cloud happens to play an integral part in how we can produce more services, better and faster for our customers. And I think it is an evolutionary necessity that we have this agility and speed.
The second is market pressures. If you look industry-wide, startups are extremely disruptive in the marketplace and are increasingly able to reach enterprise scale through leveraging the cloud’s agility. 5 to 10 years ago, an enterprise would never have had to worry much about competition from small companies. They could keep a close eye on them and watch them as they grew. Today, you have new businesses starting in the cloud with a handful of people. They aren’t burdened by legacy enterprise systems and they’re able to focus on disrupting key parts of established businesses. So if we are not ahead of those companies, we put the rest of our businesses at risk. For us, the support of senior management and a top down approach is key for the company to become more web/internet centric and agile.
GC: Continuing on this enterprise level cloud adoption trend, what do you say to someone in an enterprise organization who in your role or junior to it, who sees other larger businesses being threatened by startups gaining scale. How does he articulate the value proposition for his company as to why they should move to cloud?
DP: That’s a good question. At a holistic level, you have to start by believing in your ideas and that those ideas are in the best interest of getting the job at hand accomplished. For me, I’ve always had a vision of how technology should sync architecturally to a solution. Over the years, I’ve taken a proactive approach to driving new ideas even when they weren’t exactly welcomed. From that perspective I think you have to first believe in the technology yourself, and once you do, you have to create a vision that is compelling enough to get others to see how that vision addresses their challenges. In order to articulate the value proposition, you have to understand the challenges your are facing and then figure out if/how the cloud enables you to solve that challenge. I would suggest starting with where they see opportunities, be it agility, scalability or cost and basically build the business case so senior management is compelled to look at it. If you go back to your management team and tell them they can move something to the cloud for X% less than what they pay now, it’s hard for them to not recognize that as a significant factor.
GC: Once you have buy in for the cloud in the organization, what is the decision on what version you will move to? How do you decide how much of your infrastructure to migrate over?
DP: To me, the decision isn’t one of whether I am going all cloud or not – it’s a decision of what makes sense for the business and what considerations must be made for the legacy estate you have to manage. Especially in an enterprise that’s been around a number of years, there are a lot of historical challenges with picking up and moving something to the cloud, whether it is software that doesn’t support cloud, or its application architecture that wasn’t designed to support some of the more loosely coupled models the cloud operates in. The sense is that the mid-to-large scale enterprises that do have legacy applications won’t be able to jump completely into the cloud in the near future due to re-architecture needs or compliance issues.
When we started with the cloud, we were very clear in what we wanted to achieve as part of our strategy and our execution plan. We scoped and prioritized what we thought would drive the most value to the organization by focusing on four key areas: (1) Disaster Recovery, (2) New Product Development, (3) Non-production (QA/Stage/Dev) and (4) Legacy applications. Let’s take disaster recovery (DR) as a working example. DR, to me, is exactly what the cloud was intended for. It is the perfect opportunity to change how IT organizations run DR programs. Traditionally, they tend to look at what they have today in one data center and replicate it over to another effectively duplicating (if not more) infrastructure, software and costs. They also have to keep maintenance on both sets of equipment, keeping them up to date with code and data, and much more. In some cases, the second data center is only used when a complete failure takes place. Some people have started to think differently about DR and turned their DR sites into active/active data centers, but with that comes incredible complexity to their application architectures. When you think about the cloud, this type of situation poses an amazing opportunity. Let me try to explain – rather than having to manage another data center and taking on the full cost burden that comes with it, I can stand up a small percentage of my existing infrastructure (for data replication and code refreshes) in the cloud and scale out only in the event of a real emergency. So if you think about that alone, there is a huge cost savings to be had just on DR for those focused on things like business continuity. With that, DR was clearly the first area of consideration for us.
The second was new product development. As with any company, we evaluate ways to expand and grow our business. In some cases, we enter mature market places with existing penetration, while at other times; we are entering or creating new markets that don’t exist yet. While we understand the risks associated with these types of investments, we want to minimize our exposure and costs wherever possible. The cloud’s meter-based pricing model, enables us to take bolder steps forward and potentially take on new risks which we may not have considered in the past due to the upfront costs. The cloud will allow us to invest in successful ideas while quickly divesting in ones which don’t look as promising. .
The third area of focus is non-production environments. When you look at an enterprise, a large percentage of its infrastructure is non-production oriented, sometimes upwards of 75%. You have environments like development, QA, certification, functional testing, load testing and disaster recovery. In a lot of areas, maybe not always replicative of each other, you have a significant amount of redundant infrastructure . That is another area where we are taking a look to see how we can leverage the cloud: for non-production environments for developers and functional testing opportunities.
The fourth and final scope is our existing applications, which is probably the most complicated. With the cloud, we look at our existing legacy estate and products that to determine if there is a direct cost savings by lifting them out of our data center and into a more utility-based pricing model in the cloud. Those are the 4 core areas that guide our cloud first strategy.
GC: How does cloud play into other parts of your organization? For example, your social or mobile strategies?
DP: The cloud is driving new innovation and creating a sense of excitement in our development teams. I’m seeing teams come up with new architectures and solutions to address challenges, which are very exciting. From the business perspective, I think our partners are looking at the cloud as an enabler to help IT, help them. With respect to social or mobile, these are important areas of focus for us. We’re evaluating existing cloud based solutions to social, and think they can assist in getting to market quickly.
With regards to mobile, it’s an interesting place at the moment. Its growth is continuing to skyrocket. More and more devices and operating systems are coming to market, Windows 8 being the most recent. There is a 7” iPad, and the iPhone 5 release; new android markets come in every day. How the cloud plays into this new ecosystem will be interesting. Right now, we are looking at how we can better provide services to everyone, everywhere. So taking multi-channel distribution for example, how do we make sure we deliver the product multi-screen anytime, anywhere, whether our users are mobile, on an Apple, Droid or Windows device, at home or on a traditional desktop? We see the cloud as a potential option to help answer this question.
By Jake Gardner