The era of Big Data has ushered in a veritable gold-rush mentality across many industries, promising to solve previously intractable problems with new insight divined from vast oceans of data. Use cases abound: online retailers better understand customers’ preferences, the World Health Organization better understands the spread of infectious diseases, life science and medical device firms use genomic data to engineer better treatments, climate scientists predict the impacts of climate change, and social scientists isolate the root causes of crime and poverty. Indeed with so much opportunity for social and economic good, it is no longer a stretch to say Big Data is changing the world. It’s time to add data security to this “good” list of use cases!
The secret sauce making all this goodness possible is not simply the ability to process greater volumes, velocities and varieties of data more cheaply using technologies like Hadoop. This is crucial, but not sufficient. The real value is what can be done with this data, via the “smarts” we can build on top of it. The last few years have seen an explosion of innovation in “data intelligence” capabilities, including:
- The ability to detect, discover and classify patterns and relationships in seemingly disparate datasets
- Near-real-time analytics to identify new insights and predict downstream outcomes
- Machine Learning, the ability to learn and fine-tune from previous results
Moreover, we have seen the emergence of new data-centric roles and responsibilities: Data Stewards ensure accuracy and relevance of data (else “garbage in, garbage out”), Data Scientists harness the power of these data intelligence capabilities to produce valuable insights, Chief Data and Analytics Officers ensure the whole organization gets value from its data and benefits from these insights, and, of course, Data Security Practitioners ensure that data is being used safely and securely.
“What does this have to do with data security?” you may ask. There is the obvious use case of applying security policies to the vast data lakes being established, and most data security vendors including Protegrity either have shipping product or a credible strategy for achieving this. But regular readers of this blog know that we go beyond the obvious here! The more interesting question is how to harness these data intelligence capabilities to better solve the data-centric security problem itself.
Indeed, recent trends in the data security market are pointing towards these capabilities becoming not only relevant to the data-centric security problem, but critical to solving it in the future.
Trend #1: From Silo-First to Data-First
The biggest trend is that of moving from silo-first to data-first security. As data flows from database to database, app to app, and on-premise to cloud and back, the focus must be on protecting various sensitive data elements regardless of where they reside. This raises critical questions such as “How do I know where my sensitive data is?”, “How can I detect when new sensitive data comes into my jurisdiction (including wherever it doesn’t belong)?”, and “How can I tell where my sensitive data is coming from and where it’s going?”
As one’s data landscape grows, and the number of possible data sources expands to hundreds and thousands, any methodology that depends primarily on human beings’ tribal knowledge is bound to fail. Like asymmetric warfare, even if you find and protect 99 percent of your sensitive data the remaining 1 percent is what can be breached. Data discovery and classification can help with this and by pointing such tools at any relevant data source (and/or having crawlers go through the network looking for such data sources), it’s possible to build a more reliable data catalog against which a security practitioner can define and manage security policies.
Trend #2: Everybody is a Data User: The Emerging Data Democracy
There has been an explosion in the number of end users of data, including from human users (employees as well as external customers and partners) and applications (again, both internal apps as well as external cloud and/or mobile apps). Big Data enables more problems to be solved in a more data-driven way, which means more users trying to satisfy more use cases with more data. Indeed, nearly everybody is a data analyst now, using data to be more effective in their everyday lives. This emerging “data democracy” raises the challenge of how to empower these users to be smarter and more effective, and to do so safely and securely, preventing unsafe behavior without affecting legitimate use of data.
Technologies such as digital rights management and multi-factor authentication play a role here, but aren’t sufficient. Also needed is the intelligence to recognize valid users and user behaviors at a very fine grain, including not only the end-user but also the end-user context such as client device, location, time of day, volume and relevance of data sets being accessed, and so forth. Any solution that depends on manually defining policies for each such use case won’t be able to keep up. Instead one should also leverage analytics, based on up-to-date snapshots of sensitive data and valid users and use cases, to determine whether a given usage pattern meets existing policies and/or alert the right data security practitioner as needed.
Trend #3: The Data Landscape’s Increasing Pace of Change
New data is always coming into the organization, new use cases are constantly emerging, and new data architectures like enterprise data lakes are only accelerating such change. Additionally, most organizations’ user base is in constant flux, such as new hires, new roles, new client devices including BYOD, and new data sets that existing users may be interested in — some of those users are even creating and sharing their own data sets.
Data usage patterns are constantly evolving, challenging the data security practitioner to empower innovation without compromising security. Again, any solution depending on manual approvals and change-management won’t keep up. Instead one needs machine learning, the ability for the solution to recognize variances as being “within policy” or not, proactively alert the practitioner, and learn from their feedback. This ongoing self-tuning is the only way to keep up with the ever-increasing pace of change in an ever-more-complex data landscape.
Data Intelligence is Not a Silver Bullet
I will close with a cautionary note. The vendor community is just now starting to explore how data-intelligence technologies can be applied to solving the data-centric security challenge in a Big Data empowered world. Stay tuned for exciting news from Protegrity and other vendors in this area, but as you hear about such news, keep in mind the criteria of how to do data security right. The wrong way to do this, as alluded to in my previous blog, is to just throw this technology at the tired old silo-centric ways of doing data security. Throwing more smarts at a flawed strategy won’t change the flawed nature of that strategy. If anything, one will simply fail even more spectacularly than before. One must start with the right strategy by answering these questions:
- Are you looking at data first, not its silos?
- What is the value of that data, in business terms, and the costs and risks of not securing it properly?
- Are you first seeking to understand how that data is being used, then defining policies that best empower that usage, safely?
- Are you establishing a competency that can define, apply and measure these policies in a repeatable way, with security-practitioner roles and responsibilities empowered to achieve this across the whole organization?
- Do you have the flexibility to apply a variety of protection methods (encrypt, tokenize, mask, etc.) in a variety of ways (data-at-rest vs data-in-motion), depending on the data and how it’s being used?
Within this context, data-intelligence technologies can greatly enhance the ability of those practitioners to be successful. Outside, it is just yet another technology thrown at the problem and unable to solve it.
Bottom line: The promise of data intelligence applies to data security just as much as any other data-driven use case. With the right strategy and competencies in place, it can help maintain a strong data security posture in spite of ever-increasing data complexity across silos, ever-increasing users and use cases, and ever-increasing pace of change.
Stay tuned for more exciting innovation in this area!
The post Data Intelligence Meets Data Security appeared first on www.protegrity.com.