In association withKyndryl
In any enterprise, digital transformation is not only a technology transformation but enables business transformation itself, driving new products, solutions and innovations. Having an efficient data strategy is critical to any successful digital transformation but requires careful investment into both people and systems.
“To achieve that goal, availability of good data, of the right data, and availability of that to the right people and systems is very, very critical. So that forms the data strategy for any enterprise today,” says chief architect for data and AI services at Kyndryl, Sundar Shanmugam.
Getting the most out of digital transformation investments means evaluating and optimizing agility throughout an enterprise to drive actionable insights, says Shanmugam. A strong data governance framework also goes a long way in keeping data high-quality. Often data governance primarily serves regulatory requirements. But truly effective data governance is holistic, he adds. Data usage, regulations, and the data itself are constantly evolving within an enterprise, effectively making data governance a continuous process.
Although tech teams are often dictating how data should be managed and used, Shanmugam says everyone across the enterprise, including leaders and decision makers, should be data literate.
“End of the day, the people are the ones who design the systems and who develop the systems that consume the data, so the right investment on literacy is paramount in that aspect,” says Shanmugam.
Another key component to digital transformation lies in maximizing investments across business units. The combination of software development and operations to form devOps, AI and machine learning to form MLOps, and finance and operations to form finOps all fall under the broader umbrella of XOps that categorize these merging of IT disciplines with business operations. XOps all come together to deliver value in the most efficient way with each combination focused on maximizing automation, reusability, and agility, says Shanmugam.
“As we say, necessity is the mother of innovation, so that necessity can continuously change,” says Shanmugam. “At the core of that, if we keep that data proper [condition], then it can expand the horizons, not just internally, but even for the other external requirements and use cases.”
This episode of Business Lab is produced in association with Kyndryl.
Top Trends in Data and Analytics for 2021, Gartner, February 16, 2021
EmTech Digital 2022 session, Business-ready data holds the key to AI democratization, presented by Kyndryl
Laurel Ruma: From MIT Technology Review, I'm Laurel Ruma, and this is Business Lab, the show that helps business leaders make sense of new technologies coming out of the lab and into the marketplace. Our topic is how to build a data strategy across the organization.
To be effective, data needs to go beyond the tech teams and into the hands of the decision makers. And although that journey can be a challenge within the enterprise to finance data and machine learning operations, there are clear opportunities.
Two words for you. Maximizing data.
My guest is Sundar Shanmugam, the chief architect for data and AI services at Kyndryl.
This podcast is sponsored by Kyndryl.
Sundar Shanmugam: Hey Laurel.
Laurel: So, having a data strategy is critical to a company's digital transformation, but it requires investment in people and systems. What are some of the key best practices to get the most out of that investment?
Sundar: Digital transformation projects in any enterprise is not just a technology transformation. It aids and enables the business transformation itself, irrespective of the industry where the digital transformation is happening. In that aspect, it drives new products, solutions, and a majority of the innovations in that organization. If that is the case, then digital transformations should be able to provide the insights which are actionable and drive the intelligence that is required for the business and the critical decisions, especially, and it should happen on demand, which means that in today's world, in real time. To achieve that goal, availability of good data, of the right data, and availability of that to the right people and systems is very, very critical. So that forms the data strategy for any enterprise today.
When we look at it from that aspect, then my recommendations would be primarily around one, to look at the agility of people and systems who can adapt to continuous changes and have a strong framework for not only technology, but also for the mindset behavioral changes that are required to serve good insights, even in future. Two, have a strong governance and control for new data so that it stays always relevant and of high quality, which is also good quality. Three, keep your data foundations simple, clean and error free. That makes the data strategy fresh and enables innovations for that. Those are the three best practices that I would recommend for any data strategy for the maximum investment.
Laurel: Great. So going back to that one, that clean, usable data, that's certainly one part of the story and it's one of the three main best practices, but metadata, how data is described, is also critical. Can you explain why metadata is such an important part of data strategy?
Sundar: So, we spoke about the data strategy and the data strategy is for long term and that should serve the business transformation and the innovations for the business itself. So, if that has to happen, then it is very important to record the details about the data itself, which is what we call it as metadata. When the metadata records the information about the data, it serves multiple purposes. One, as the data evolves because the data is not constant in today's world, so that data is continuously evolving and the usage of the data is also continuously changing. To meet those continuous changes, it is very important to know where the data has originated, what changes it has gone through, so if you have to change some of the attributes, we know what we are doing and make it reusable whenever we need the data.
It is also important to keep this metadata consistent across the lifecycle of the data–from where the data is being produced all the way to where it is being consumed, so that the information about the data itself is consistent. That would further enable not only the clean and usable data, but it also makes the data more self-consumable and more consistent at the hands of the users. That's why the metadata is a big part of any data strategy, especially in the evolving industries where the data is critical.
Laurel: Yeah, you've touched on some very important aspects here. Mostly that data governance is quite critical and it needs to be incorporated into the technology processes and practices across the entire enterprise. What are some best practices for company leaders establishing or restructuring a data governance framework?
Sundar: In my experience, being an architect in the past and managing and providing consulting for a lot of my customers, data governance is being looked at primarily to serve the regulatory requirements in the past. So, it used to be a standalone process level, but for any effective data governance there, it should be a holistic process. It should be done right from the source of the data all the way to the consumption of feedback. That is one of the key best practices that we recommend to all of our customers. Also, data governance is a continuous process. It is not that, "Okay. I looked at the requirements of the data today," whether it is regulatory requirement or the consumption requirements, "And I devised a plan for that and I can take risk now." No.
So the data governance is a continuous process. The requirements of data continuously change. The usage of the data continuously changes. Regulations are continuously changing. So the data governance process and revisiting that is also very important and a complete understanding of what is happening, what has changed, why it is changed, when it has changed and keeping a record of that is also very important. That's why the data governance framework should have a holistic process. It's not a siloed process and it should be continuously revisited, and it is continuously tracked as well.
Laurel: And as you mentioned earlier, people are definitely part of this process and strategy as well. How do you think about data literacy as a critical skill that everyone needs to have across the organization outside of the tech teams? How should executives start thinking about preparing and ensuring everyone has those right skills to consume data?
Sundar: So, data is the “new oil” that is being fed everywhere. If data is a new oil, the understanding of how to use it, where to use that data becomes very, very crucial. How to use it and where to use it forms the major part of data literacy in any organization. Also, if we have to use any given data, then we should also know where the data is available. So, data literacy is addressed at two levels. One, about providing the information on what is the data that is available, how good that data is that is available, how to access that data, how to process that data. And the second one is, especially in today's world, the data also has many constraints. It is very critical and it has a lot of sensitive information. The line between the sensitive information and the data that can be consumed easily is very thin in today's world.
If that is the case, then the literacy of what data that we are processing and how sensitive it is, what we want to use with that, that literacy of that information is also very critical. So when the executives plan for data literacy programs in their organizations, it is also important to make sure that it's not only about the data usage, but also what is the usage of the data and what is the outcome of the data? So, that's why data literacy and the investment of data literacy on people becomes very critical. End of the day, the people are the ones who design the systems and who develop the systems that consume the data, so the right investment on literacy is paramount in that aspect.
Laurel: So, those are very important parts about data literacy, especially across the entire organization, but we've also seen that another part of digital transformation is streamlining and maximizing investments in operations across business units. For example, years ago, tech teams did this by combining software development and operations to create devOps, which allowed for more agile and data-focused ways of working. The research firm, Gartner, argues that this philosophy can also be applied to other areas of the business, including artificial intelligence and machine learning to create MLOps, data to create dataOps, and finance to create finOps, so finance and operations. As a whole, these can be bundled into one single term called XOps. It's an interesting way to take various parts of the business and bring it all together under an umbrella of operations. What value can XOps bring to an organization as a whole?
Sundar: Yes, as you rightly said, Laurel, XOps is an umbrella that brings in various operations that drives innovation through the technology to address the business requirements to take the business to the next level. Having said that, all the three operations, for example, that you have mentioned, whether it is devOps, dataOps, MLOps, or even finOps, the fourth one, everywhere, the common denominator's operations and the requirement for that operations is to deliver value in a most efficient way.
So what we learned from devOps is managing versus developing a product, how to combine them and extract that efficiency. The same principles are taken into machine learning operations and data operations. Again, from the technology perspective, the common factor there is automation and continuous reusability of the processes to make that entire operation efficient. That's why Gartner has combined all three and they call it XOps, so you can look at it like a Venn diagram of three different operations, which pivoted around automation and reusability with agility.
To run all these things, especially with the data and machine learning, the data is continuously increasing, the volume is continuously increasing. Similarly, to process that, the machine learning activities and the resources used by these processes is also continuously increasing. That is where the finOps becomes very critical to keep all these things in check. That's why they are all combined under one umbrella, and they all come together to deliver the value in a more efficient way. That's the huge difference that XOps brings to an organization, which is more agile and more cost effective for the data and EA operations that they perform in their organizations.
Laurel: So how can XOps help give visibility to the success of cloud computing adoption? How can finance help tech make better decisions for the enterprise in that kind of relationship?
Sundar: As we have seen before, the finOps component of XOps is what makes the financial controls on the entire operations of data and machine learning more robust. To look at it from a different perspective, the adoption for cloud computing itself was primarily driven for cost advantage when it started 10, 15 years back with infrastructure as a service. So, if that is the prime base of cloud computing adoption, in today's world, when it is being used for data management and machine learning model management, it becomes highly imperative to continuously monitor the data consumption and the resources that are being consumed by the processes involved in the machine learning.
So, finOps looks at cloud computing, the data that is being consumed, the models, and even to the extent of whether the models are decaying, how much resources they are consuming, etc. By bringing all this together, they give very clear visibility to whether they're able to achieve the goals of cloud computing adoption in the first place. The service of EA projects and the outcomes from those projects. That's how finOps, which is part of the XOps, is helping to measure the success of cloud compute.
Laurel: So if XOps helps prove that success for cloud computing adoption, why would this be so critical for executives who are probably looking for that ROI, return on investment, for their extensive cloud computing investment?
Sundar: Yes, the cloud computing investment is part of the digital transformation and it should deliver the business outcomes in terms of the value from the innovation projects where the EA and data projects are primarily involved. So in that aspect, the value realization from the data and the eventual EA projects makes the outcome from the cloud computing investments realized. If the data is not of a good quality, and if we are not able to measure how the data and machine learning function, server and performing, including their own resource consumption as well, then it is very difficult to map it to the outcomes that derive from that. That's where the outcome of cloud computing really depends on the finOps and the data and EA projects itself.
Laurel: So, when we think about data and XOps and access to data across the organization, we also need to really think about security as well. How can this holistic view of data with XOps help an organization keep their data secure, but also enable people to make those critical decisions with the data in an easy to access real-time way?
Sundar: Yes, so that data fits into the EA and the eventual insights from dev, so the operations related to data to keep the data clean and usable also means that the data should be available to the right people in the right format for dev operations. By bringing the data operations and machine learning operations together under the umbrella of XOps, it keeps the data closer to where the data is going to be processed. The profiles for this data consumption can be maintained in one single place. That way, the data access, as well as the consumption of that to easily process them and access them, becomes much more streamlined and efficient. That's where the XOps helps the data to be more secure, at the same time, easily accessible, not just for one experiment or one project, but in a continuous manner.
Laurel: You mentioned earlier that having a good data strategy helps with innovation and the adoption of emerging technologies. How do you see that playing out at various companies? Where is that innovation possible?
Sundar: Oh, the innovation around data and EA in today's world is everywhere. There's no limitation or boundary for the innovations that are available for the enterprises today. There are examples like in the automotive industry when the data strategy is being formed, the data strategy is primarily revolving around how to make it more usable, more in real time within the organization. But when the same data has extended its boundaries for requirements and uses like the connected car, the data expands its boundaries. The data is not just usable for their own consumption anymore. The automotive companies are able to monetize the data by extending it to others. In one of the projects that we have delivered, the insurance companies are getting information about the vehicle performances, how the driver is using that vehicle, what are the different demographics that are being used? The data which originally originated from the automotive industry is now being consumed by the insurance industry in this case.
And that innovation has moved from the automotive industry to the insurance industry. So that way, the data strategy, if it is correct, and if the data is in a good state, and if an enterprise uses proper data management techniques to keep the data always relevant, then the innovations can and will continuously grow as the time progresses and as the requirements continuously evolve their way. As we say, necessity is the mother of innovation, so that necessity can continuously change. At the core of that, if we keep that data in proper [condition], then it can expand the horizons, not just internally, but even for the other external requirements and use cases. There's no limit or boundary for the innovations with the data that we can see in this case here.
Laurel: It's certainly very exciting, lots of opportunities with that innovation. So how are you looking at the next three to five years? What's the data landscape and the way we work with data, how is that going to evolve?
Sundar: There are again, immense amounts of possibilities and opportunities there, especially if you think about three to five years. From a technology aspect, there are various things happening like Meta. When we speak about Meta or augmented reality, again, the amount of data that is going to be generated is humongous and the usage of the data and what we can do to develop insights and to make a huge change in the way different industries function is also very high. That is on the technology side. At the same time, we look at how it touches and impacts the people and environment like sustainability. So sustainability is a big thing today and data and insights power sustainability. It's not just to measure sustainability, but even to forecast, to devise new strategies on sustainability.
So that way, in my view, the data is going to grow a lot, and that's where we are seeing the data mesh, which is a technology or technological principle that makes the data ownership lie with the business units because they know the data better and they can share it with others based on the own regulations, and the consumers can also use them more efficiently. So the data is going to be much more powerful. They're going to have a great time coming up with new insights in the next three to five years. So the metaverse and augmented reality on the technology side, sustainability on the world environment and societal side, and data mesh in the data architecture principle sites. These are the three big things that are evolving in my view in the next three to five years.
Laurel: Excellent. Sundar, thank you so much for joining us today in the Business Lab.
Sundar: It's really great speaking to you, Laurel. Have a great day.
Laurel Ruma: That was Sundar Shanmugam, the Chief Architect for Data and AI Services at Kyndryl, who I spoke with from Cambridge, Massachusetts, the home of MIT and MIT Technology Review overlooking the Charles River.
That's it for this episode of Business Lab, I'm your host, Laurel Ruma. I'm the Director of Insights, the Custom Publishing Division of MIT Technology Review. We were founded in 1899 at Massachusetts Institute of Technology, and you can find us in print, on the web, and at events each year around the world. For more information about us and the show, please check out our website at technologyreview.com.
This show is available wherever you get your podcasts. If you enjoyed this episode, we hope you'll take a moment to rate and review us. Business Lab is a production of MIT Technology Review. This episode was produced by Collective Next. Thanks for listening.
This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff.
Humans and technology
The worst technology failures of 2023
The Titan submersible, lab-grown chicken, and GM’s wayward Cruise robotaxis made our annual list of the worst in tech.
Unlocking the power of sustainability
A comprehensive sustainability effort embraces technology, shifting from risk reduction to innovation opportunity.
Drive innovation with a tech culture of ‘connect, learn, and apply’
Offer an ongoing curriculum that aligns to strategic priorities and the latest technology to drive innovation, productivity, and social-good efforts.
People are worried that AI will take everyone’s jobs. We’ve been here before.
In a 1938 article, MIT’s president argued that technical progress didn’t mean fewer jobs. He’s still right.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.