Let your business objectives drive cloud implementation
A Q&A with Adam Wray, CEO and President, Basho Technologies
Published March 18, 2015
There is no one solution when it comes to cloud services and cloud implementation. “It’s not just about technology. It’s about business objectives,” says Adam Wray, CEO and President of Basho Technologies.
There is tremendous opportunity available through companies like IBM, Microsoft, Amazon, and others. However, organizations should focus on the business value that they want to drive out of that workload, rather than start with the technology.
Wray has been in the enterprise cloud technology market for 20 years, with operational leadership roles at Amazon, Akamai Technologies, and Limelight Networks under this belt. He most recently served as CEO and president of public cloud provider Tier 3, which was sold to CenturyLink in 2013.
In this Q&A, Wray talks about the challenges that surround unstructured data and how organizations should approach the cloud conversation.
Insights Magazine: What are some of the biggest challenges companies face with the onslaught of unstructured data?
Adam Wray: Right now clients are faced with a mantra within corporate America saying, ‘You should keep all your data; you should be looking for more data to collect.’ Part of that stems from two major drivers within the unit economics. One of those drivers—data as a strategic enabler—has dramatically increased over the last three to five years. Conceptually at the executive or C-suite level, there is a very different point of view as looking at data as potentially the next enabler for new monetization efforts or differentiation with your competitors. The other major unit economic driver going on is the hardware component of storage, which has dramatically dropped in price leading many across the enterprise to adopt the mentality of keeping every piece of data regardless of whether you think it has any value.
These two major unit economic drivers are causing companies and executives to have to look at unstructured data very differently. When they start to say, ‘Well, keep everything,’ the next question is, ‘Ok, how do we do that? How do we potentially position that for value later?’ And then they look across unstructured data and say, ‘I have no clue what NoSQL provider to use. Should I use NoSQL, should I use Hadoop? What are even the subtle differences? Oh, and by the way, my entire internal teams only understand relational databases. So I have what seems like hundreds of choices to go forward, but I don’t have enough skill set internally to support it.’ That’s the challenge.
IM: How should companies best evaluate what is right for them? Where should they start?
Wray: Companies are going to have to cultivate relationships and cultivate talent internally to understand where they should start to make bets on unstructured data. In my opinion, as one who ran many P&Ls and ran a cloud services company and now runs an unstructured data company in Basho, the best place to start is to look at the workloads themselves. Workloads are active; they should be responsible for delivering real value relative to where they are located, whether you are talking some kind of supply chain, some type of e-commerce shopping cart, or you're talking some type of log or file system, etc. So if you start to ask, ‘This is my workload, this is what I’m trying to do; what are the possibilities of other things?’ Then you start to get a question of, ‘Well what might I need to collect or what have I not thought about collecting before?’ A perfect example of this in our own client base is bet365, one of the largest gambling sites in all of Europe within the gaming sector.
bet365 started to recognize that if they could collect more data in real-time sessions on their users who are making bets, they might be able to give them new options in which they could actually engage in more exciting ways to spread around the wealth, so to speak. One of the things they did when they went to an unstructured database, which happened to be ours—which is Riak—is they were able to allow customers to change their bet midstream. They couldn’t do that in a structured world, but in an unstructured world they could push that agenda and therefore increase the engagement level and increase the possibility that [customers] might extent their bet or put more money on the table—more fun for the client; more opportunity for bet365.
IM: What's your advice for executives who are evaluating database solutions?
Wray: It’s tough to assess what you should buy if you don’t know what your workload needs are. If you think about how the unstructured database market is—many analysts estimate it at $3 to $5 billion, and that includes everything within Hadoop, which is all archival and repository, and NoSQL, which is all active workloads for real-time type results. This goes against the $40 billion relational database market, which is a well-known 30-year industry. The usual approach when you are trying to understand a new technology is to buy something and hope it covers a lot of things in terms of features and functionality. I’ve always been a proponent of the approach that it’s better to ask the question, ‘What is your business need; what is the workload itself trying to do?,’ and then you can start to ask the question, ‘Does that database accomplish them?’
For example, if your key characteristic is availability at all cost, then I can promise you if you looked across the industry you’re going to find that on the NoSQL side, nine-tenths of the solution sets will have a certain degree of failure that will not be acceptable to that level of workload need. So you can start to use parameters like that to ask the question, ‘What am I after at my core, and how do I think I might expand from there?’ to narrow down from 100 to a couple.
IM: When should companies introduce cloud into the conversation and how can they determine which cloud configuration—public, private, hybrid—is best for their needs?
Wray: There is usually some type of business driver that causes companies to reassess how they might go about this. We did a lot of studies when we were building hybrid clouds back at Tier 3, and we always found that agility is the real driver. A company has, say, some sort of legacy workload that sat within a core location, on-premise environment or some managed service provider, and now they are trying to ask the question, ‘Now I think I’m going to do half a dozen more things and I don’t want to necessarily have to build out that infrastructure, so how do I leverage an IBM SoftLayer, an Azure, a AWS, in extending that workload out?’ When you start to do that, you start to have to do demarcation of the workload and ask the question, ‘How much am I willing to let go of to reside in another environment?’ This is more from an enterprise perspective, if it’s being planned from a business level. There is also the shadow IT or grassroots level, where people will just take stuff and start to connect it back to the core across a hybrid-type environment, which might not necessarily pay attention to security protocols or internal governance, that type of thing, but it happens a lot.
However it’s approached, the one thing a lot of people have a hard time getting their head around is that if you accept that you’re going to have a portion of your workload residing in two different locations, most often the latency involved in that is not thought through. The fact that data has real weight, i.e., data gravity—Dave McCrory’s penned term for the fact that applications will stay closer to wherever the data actually resides—and you need to factor that in to your end use and into the experience you are trying to deliver. For example, if you go to a website and it’s pulling from a database that is actually in two different facilities that are geographically dispersed and it’s a long term pull, you might sit on a website for 10 plus seconds. A lot of users will time out on that type of experience.
IM: What kind of workloads are best suited for a hybrid configuration?
Wray: We’ve all lived in technology for so long we have a tendency to always drive this from a technical point of view, but I find that now that we are at a point of maturation and [it shows] value in hybrid in saying, ‘I don’t need to own everything; I don’t need to control every component I have.’ I personally think business agility is a much more valuable thing than ownership. The challenge that I think a lot of executives have to get their head around is, do they have strong enough orchestration tools, strong enough governance in place, to be willing to allow their internal clients to let the workloads go based on business needs. If you have good orchestration tools—and I’m thinking top-down hierarchical point of view for governance. People often think about their workload only in terms of orchestration as turning on a bunch of servers or turning down servers in, let’s say a Softlayer, versus me putting new machines in a rack in our core location facility. That is a portion of orchestration, but from my point of view if you are focused on the workload and hybrid cloud, you should ask the simple question, ‘Do I have rules in place that filter from a top-down perspective on what our policies are, what our security measures are, what any regulatory components might be, what's allowed to go outside from a data construct, what ambitions or commitments do we have to our clients?’ All these things should help to drive and be more flexible in regards to using some kind of automated tool set.
It might end up costing more to have it run somewhere else, but if I get more ability to execute more quickly on my objectives, what's that matter? I think way too many people think cloud and hybrid are about cost savings; I think that’s very myopic. I think it’s very short-sighted to think that is the real value. Amazon has been training people for years that cloud is about saving money, and the reality is that is the short-term value. The operating expense savings is finite. What's actually more strategically valuable is the agility to be able to ramp up 300 servers to run a marketing campaign on the fly based on an opportunity you saw and you want to act on. You never would have been able to physically do that seven years ago. So what is the business value of that? It’s a lot. Who cares if you spend a lot of money, as long as you're getting the return on the equity invested. Cost savings shouldn’t be the driver; the driver should be, ‘What am I trying to accomplish? Do I get the business agility?’ And then you are looking at that investment for moving to some hybrid-type environment as a component percent of your enabler.
IM: For people apprehensive to move to the cloud, is big data implementation possible without it?
Wray: The apprehension is still there for obvious reasons; we’re going through a transformative decade. Data has real weight, so it can be a real inhibiter, If you suddenly decide to move large computations to a hybrid environment you might have to then move all those applications that are co-dependent on it there because of latency issues—once again there are business dynamics that have to be factored. But if I was running a traditional P&L for a large company versus a technology company I would be insistent on taking the concept of business agility and applying it against every paradigm that we have internal control over so that I could ask the question, ‘Do we get strategic advantage by putting this into some type of hybrid or public cloud or SaaS provider?’ If the answer was yes, I would be driving people to do it.