Q&A: Modernizing Data Management: What Works and What Doesn’t
If you're managing today's new data realities and the push for digital transformation using yesterday's data management techniques, you're headed for trouble. Aerospike's Lenley Hensarling offers suggestions for what to do and how to get started.
- By James E. Powell
- June 1, 2021
With the growing importance of real-time data, the move to the cloud, and customer demands, it's no wonder that data management policies and processes have to change to keep up. What trends are making an impact on enterprise data management? Which techniques work and which don't? How does an enterprise get started and what trends will be important in the next few months? Lenley Hensarling, chief strategy officer at Aerospike, shares his insights and predictions.
Upside: What's driving enterprises to modernize their data management?
Lenley Hensarling: In the wake of the pandemic, enterprises are doing a digital "level-up" and accelerating digital transformation. Most digital transformation is data intensive. It requires enterprises to collect, store, move, integrate, and process large amounts of data to deliver the near-instant, accurate experiences users demand. They're finding out that poor data management strategies can cause their digital transformation efforts to stall or even fail.
How are enterprises going about this modernization? What are they focusing on during this modernization project?
The best approach enterprises can take is to organize real-time data around processes. Digital modernization is based on analyzing real-time data closer to where the actual processes are happening -- closer to the customer, the device, or the transaction. Enterprises have to think carefully about the amounts of real-time data and the velocity of interactions within the processes that will be consuming the data.
Customers are now equipped to engage with an organization's processes 24/7, and they expect that everything will be processed and tracked in real time and open to them to query status. For example, recommendation engines, identity management, fraud prevention, and digital wallets are some of the programs interacting in real time with an organization's customers.
What have they tried that hasn't worked? What does work, and why are these approaches or practices useful?
Digital transformation or modernization of enterprise systems is primarily centered on opening existing processes to a wide range of customers and partners. The first knee-jerk reaction is attempting to "skin" the transactions from back-office systems with a friendlier user interface. The next step is to put a cache in front of the databases that support the back office to deal with the non-linearity of the Internet business. You can't control things when a pandemic hits and your customers try to reach out and purchase additional parts to buffer their supply chain risks. This motion is mainly about Band-Aiding the sore points.
Ultimately, enterprises need to make the journey to a highly scalable real-time system driven by data. By "twinning" the existing batch-oriented back-office systems with a real-time intraday (or intra-time-period) system, they can start to use their cache data store as the real-time database system of record. They now have a scalable, real-time system that supports digital business's pace while leveraging the existing back-office infrastructure to deal with regulatory and reporting requirements.
What's the biggest mistake organizations make when modernizing data management? How can they avoid this mistake?
A big mistake organizations make is not anticipating the volume of interactions with real-time data, which doesn't allow them to reach the required scale and performance. We call this "dealing with the non-linearity of the Internet." We're talking real-time data volumes in the terabytes and transaction volumes in the millions and even billions of interactions.
It is not only transactions but also customer status queries. Customers and partners check on things all the time now that they have a window into a business as delivered by their mobile devices. Today's customers expect organizations to know everything about the interactions they have had with the business. For example, customers expect an e-commerce company to provide a friction-free shopping experience by delivering the right offer, price, and purchase approval in the fastest time possible and then knowing when -- at any given moment in the future -- the delivery is going to happen.
Why aren't more enterprises updating their data management strategy? What's getting in their way?
Data management strategy is less like navigating highways and more like paddling through rapids. The actual path is constantly changing in real time, and organizations have to be reacting and looking ahead and adjusting their course continuously, or they won't make it through. In the past, enterprises thought about managing strategy and technology at points in time. That does not work anymore. It takes continuous awareness and assessment and an agile response.
Evolution is constant, and systems are no longer static. New DevOps models allow enterprises to implement and deliver this way, but management has to be in that same 24/7/365 world of constant assessment and adjustment.
How long does a data management overhaul take? What skill sets are needed? Who should be involved in the project?
It takes a long time. However, the value can flow right from the start. It takes focus. A lot of this is about determining what will deliver the highest business returns and then relentlessly moving forward through iterations that provide value at each step. Business leaders should be involved, even driving the efforts from the beginning.
New data technologies will be involved, from non-relational distributed databases to new data distribution and processing models like Kafka and Spark. Enterprises shouldn't slight the acquisition of talent or the training of technical personnel. We see our customers get returns in months, not years, and continue to move forward continuously, expanding from success to success.
How does an enterprise know that its current strategy is no longer working, and how often should an enterprise revisit its data management strategy even if there are no obvious problems?
Business is competitive, and enterprises must have an open-eyed assessment of their performance amongst what they perceive as their peer group. They also need to look at where the disruptors are and what approaches they are taking. Enterprises need to start with where they need to go with their business model, and what technology they will need for that becomes apparent. They can't shy away from getting outside expertise or advice. It is time and money well spent. Just because things are "working" doesn't mean that the business is not at risk.
What trends do you see ahead for data management? What should enterprises be paying attention to over the next 6 to 12 months?
One trend is enterprises will become far more savvy and deliberate in deciding which workloads should be run in which clouds -- public, private, hybrid, or multiple. They'll start designing for a specific cloud depending on if they need heavy compute, a lot of storage, or network bandwidth -- or maybe all of it.
Another trend is enterprises will implement digital augmentation strategies. Not every system or business process can -- or should -- be ripped entirely out and reinvented. Adding modern technologies around existing processes and systems can often dramatically speed transformation to something familiar but far more efficient.
A third trend is that when Google phases out third-party cookies in Chrome, enterprises will have to reexamine how they gather customer intent -- and process a new, more complicated mix of hundreds of data sources to know their customers' digital behaviors.
Finally, customers' expectations and the digital world's demands are mandating real-time responses to business events.
Editor's note: Lenley Hensarling is the chief strategy officer at Aerospike, creator of next-generation, real-time NoSQL data solutions. Lenley has more than 30 years of experience in engineering management, product management, and operational management at both startups and large successful software companies. Lenley previously held executive positions at Novell, Enterworks, JD Edwards, EnterpriseDB, and Oracle. You can contact Lenley via email.