If data sits on a desk somewhere and is not being used, it’s an opportunity wasted
Sanjay Mirchandani believes IT has to take the lead in adding value to the business in the form of big data “addictive analytics.” Mirchandani is Chief Information Officer and COO, Global Centers of Excellence, at EMC Corporation. He has been recognized as one of Computerworld’s Premier 100 IT Leaders and Boston Business Journal’s CIOs of the Year. The following is an edited transcript of our recent phone conversation.
What would you say to a CIO who dismisses big data as just another buzzword?
I would say that for too long we have been trying to manage down information. The IT world that we have become comfortable with for many years was mostly within the enterprise, maybe connecting to some partners and customers. It was also mostly structured, basically revolving around transactional data. Today, the volume, variety, velocity and complexity of information have changed the IT landscape. These are the four things I challenge CIOs to really think about. We all know how to do structured information. But the moment you throw in unstructured and semi-structured information, life changes. This is where the value is for organizations today.
Does this also change the relationships between IT and the business?
Only IT has a complete picture of all the data in the enterprise. At the same time, IT today cannot have a monopoly on information. That changes the role and responsibilities of IT and the business. We in IT want to deliver more as a service and the business wants to consume more as a service. And IT and the business increasingly share tools and capabilities. For example, I can offer a tool like Greenplum Chorus, which is a community-based BI-data warehousing-analytics tool, where data scientists in IT work collaboratively with data scientists sitting in the business. If there’s something we can do better, we’ll take it on ourselves; if there’s something they can do better, like creating their own wrappers around the analytics, they will do it. What’s clear is that IT and the business have never been better aligned.
To move into this brave new world of big data, what’s the first step a CIO should take?
Too often we get mired in what we know best. It’s a good time to step back and look at five key foundational layers of a big data architecture. The first is sources of data, especially new sources of data such as social data, external data, and machine-to-machine conversations. Doesn’t matter which business you are in, there’s a lot of data today that is generated through sensors’ feeds. So, take a step back from the transactional data you are familiar with and take a look at other sources of data that are or need to be core to your business.
The second layer is data governance. For too long data governance has been a bad word. You need to take a fresh look at how you manage your data—if the data that is going in is incomplete or dirty, the analytics will reflect that. You should put in a data czar and a team that works on keeping the data clean, on rules and regulations and governance around who can change data sets and data structures. At EMC, we have a data governance committee which includes senior people and an expert on master data management to help us establish the right process.
The third layer is to select and implement a big data platform for your company. You need to pick one that does structured, unstructured, semi-structured data and does it with scale-out capability. Notice I’m not talking storage. Where you store it is a function of what you do with it.
Next, think about analytics tools on top of that. People usually pick what their data scientists like. I don’t think there is any religion around it, and as long as it uses standards, you are in a good place.
And then you need to think about the data scientists—which should live in the business and which should live in IT. Think also about your visualization capabilities, your training, back-end and front-end tools. And you need to put a wrapper around these moving parts so it’s easy for the business to consume. We call it “BI-as-a-Service” but you may call it anything you like. The key is to think about these five layers and build your strategy around them. And I’d say, walk before you run. But not doing it is just criminal.
Is BI-as-a-Service the only option for a CIO considering big data?
I don’t think it’s the only way, but I think that’s where people will logically end up. It fits within the general trend of more collaborative work between the business and IT, as tools and capabilities are getting better across the board. Over time the roles will shift to be more collaborative.
What are the benefits, for IT and for the business units, of your approach?
I’m going to flip the answer and say that it’s an incredible value for the company you work for. Data is an underplayed asset. For too long, IT organizations have been giving themselves a pat on the back about how little they let the data grow. The growth of data has always been considered a bad thing. Today, companies that understand big data see it as an incredible asset for them from a competitive vantage point. IT’s job is to align with the business and bring them value, bring them clean data sets, bring them fresh perspective on the company’s information assets because they are the only ones that are able to see all the information, end-to-end. And the business needs to understand that big data is a good word, not bad word. If data sits on a desk somewhere and is not being used, it’s an opportunity wasted.
That’s a great value for the company, no doubt. But what specific benefits have you seen so far or you expect from providing BI-as-a-Service?
Doing it as a service allows my IT team to focus on things that are important to them and it allows the business to do what the business likes. Companies like ours have been plagued with shadow IT for years. Parallel IT, business-managed IT, call it whatever you want, happened because IT didn’t give them what they wanted, or they didn’t like what IT gave them or they thought they can do it themselves. Shadow IT today, with much more empowered users, is a given. My approach allows it to come out of the shadows and become mainstream.
This means valuable data can now turn into a corporate asset as opposed to a fragmented data set that is hidden somewhere. The business can get answers to their questions much faster than before and they have the freedom to explore, to play in a sandbox on top of the corporate data warehouse. And we have seen a 30 percent reduction on average in the cost of delivering analytical capabilities to the business as opposed to the old model of each business doing it on their own.
When I say move the needle out, I say let IT do the data governance, management, control, compliance, and give you one version of the truth. You, as the business, knock yourself out and do the analytics you want any way you want to do it and provide competitive differentiation to the business. If you want to work with our data scientists, great; if you have your own data scientists or a third-party that helps you, fabulous. But there is only one place that you come to get the data and that’s IT.
It’s not a corporate compliance approach. It’s a value-to-the-company attitude.
Some people argue that the best place for data scientists is somewhere outside of IT. Why do you have data scientists in your IT organization?
It is not different from traditional data analysis. Every business unit typically has analysis teams focused on their domain. Our data scientists will need to work across all of the units with a raised level of sophistication and the right level of abstraction. It’s akin to an academic community—data scientists like to work on each other’s analytics, on each other’s problem set, they collaborate without much encouragement.
I’m not setting up a large data science team—I’m setting up a core data science capability and what they focus on are big data platform expertise, IT-type problems and potentially some data abstraction capabilities from a cross-functional perspective. The business focus is on the subject matter expertise around what they do. I think it works hand in glove.
Does Data Science represent a new growth opportunity and exciting career move for people in IT?
It sure is. There isn’t an abundance of people with this skill set. The BI-Data Warehousing-Analytics space is creating a lot of new opportunities for IT professionals. It’s a good time to be in IT.
It’s also a great time to transform IT to become a change agent, a catalyst on the road to more and more data-driven decisions.
IT needs to be agile. IT needs to understand that our role is to be brokers of change. And not be religious about whether we write it or buy it. I know that when I speak to my peers, other CIOs, we are all focused on transforming IT. The essence of this transformation is changing our go-to-market model because the traditional tool sets and competencies are no longer our monopoly, they are readily available to the business.
If you want to run IT as a business, you’ve got to look at the “competitive” forces. Anyone with a corporate credit card can buy whatever IT resources they need, they can buy it as a service. We understand that the transformation in how we go to market is our number one priority.
So you see IT less focused on running the infrastructure and more on helping the business find value in data, its most important asset?
For too long people have been making money talking about how to align business and IT. We are in an era where business and IT are aligned. If you have a cloud infrastructure, what does it run on, how do you run it, what is the cost of managing it, this stuff is all documented, available, and delivered as a service. Let’s not waste time on that, let’s worry about what value the business wants.
Do you think this now a generally accepted approach to managing IT?
Well, we are all grappling with it, with the new IT go-to-market model. What is the face of the new IT for the future? How does cloud change that? How do mobile devices change that? The best answers are coming from the people that are in IT today because they understand that the solutions are in new ways to deliver IT services.
What are the typical stumbling blocks for companies that want to transform their IT organizations?
Legacy systems are an impediment. If the company is slow in moving to a cloud-based, flexible infrastructure model, then they can’t change their application base. If they can’t change their application base, they can’t deliver the value of mobile computing, etc. Their people live in the past. They have one foot permanently entrenched in legacy while they are trying to flirt with new ways of doing things like cloud computing. It’s very hard on IT folks to be in this situation.
You got to make the investment to get a flexible infrastructure, you got to get out of the silo business, you got to adjust your delivery model so it’s just-in-time, you got to start looking at the apps stacks. And you got to move fast. If you spend a lot of time vacillating on which technology to use, your people get mired in the past instead of moving forward.
These new technologies give rise to a new alignment between IT and the business revolving around what is most important—the data.
Applications and the data. The usability and then the data, which is becoming the next big value layer. But you got to get the infrastructure conversation locked and loaded, you’ve got to go build it, you got to move your people through that quickly. Half-legacy, half-cloud for too long is hard on people.
Do you think that in the future we will see more companies providing to other companies big-data-as-a-service, the entire big data analytics process from collecting and cleaning the data to analysis and visualization?
There are parts of this already today but my short answer is yes. If there is value to be gained for your corporate model to do it this way, absolutely, it doesn’t have to be one of your core competencies. But using the data and getting value from the data, must be owned by the business.
Big Data Analytics is not only for retail Business Intelligence, even though that is where some of the greatest advancements are currently occurring. Big Data Analytics is also the future of Infrastructure Asset Management. Each industry has core infrastructure that must be Asset Managed over its life-cycle through predictive modeling. Big Data Analytics will evolve too rapidly (with increasing “volume, variety, velocity and complexity” of available classes of data) for any industry organization to standardize and maintain “THE” method of doing Infrastructure Asset Management through Big Data Analytics.Those wishing to take a leadership role in the Big Data Analytics required for successful Integrated Asset Management of infrastructure need to establish the standards for the “backbone” of Big Data in their industry sector… that is industry associations should establish the standards for “data governance, management, control, and compliance” through a Central Data Warehouse. Then let consultants, utilities, software companies, and academics “knock yourself out and do the analytics you want any way you want to do it and provide competitive differentiation to the businesses. If you want to work with our data scientists, great; if you have your own data scientists or a third-party that helps you, fabulous. But there is only one place that you come to get the data and that’s [the industry association’s Central Data Warehouse.]” From the CDW, infrastructure asset design and performance can be independently Validated, Verified (iV&V), and benchmarked against peers (as is done in the software sector). As a civil engineer, I believe this is the future of the engineering “standard of care” in all sectors and will change engineering practice as we know it.
Excellent and visionary article! Thanks for the additional insights.
Pingback: IT Roles changes in the new cloud era.. – oh yes, this minor aspect that we don’t pay attention to.. « Itzikr's Blog
Big Data makes the IT and business to think differently. Unstructured data gives benefits to your business. Good article!
Pingback: The Big Data Meme: 5 Scenarios for IT | What's The Big Data?
Pingback: The Big Data Interview: Sanjay Mirchandani, CIO, EMC | What’s The Big Data? – Scientific Forefront
Pingback: Buy The Kool-Aid! | Big Data