There is exponentially more data and more users of data in digital enterprises. That means there must be a fundamental difference in how that data is stored and managed.
In the previous blog we discussed the concept of Datanomics, the economics of data, and how it has radically evolved with the digital transformation of an enterprise. Value of data is dependent on frequency and speed of access needed to deliver business requirements. Digital enterprises operate with radically different Datanomics than conventional physical businesses. Here, digital information is the business.
Yes, that means that there is exponentially more data to store and manage. But it also means a fundamental difference in how that data needs to be stored and managed.
In the late 1990s, Netflix and Rosetta Stone were physical businesses, delivering entertainment and education, respectively, and shipping DVD/CD media to their customers. A customer-facing website, applications for customer relationship management (CRM), enterprise resource planning (ERP) and billing made up for most of the data being created and processed. Data management was based on the well-worn model of exponential decay with age. Operations teams established processes to align with this model, and storage vendors offered solutions to neatly align with this model.
In a traditional physical business, data has the highest value when it is created and frequently accessed by the production applications. High performance is often required at this stage. The emergence of purpose-built applications for Oracle database or SAP HANA are responses to delivering on this high-value data infrastructure.
As custodians of data, operations teams use tools to back up this high-value production data to protect from local failure. More tools are used for disaster recovery to protect from site failures and even more tools for data archival to meet the regulatory compliance requirements. Frequency and speed of access to data exponentially decreases with time, and storage vendors developed a portfolio of products to align with this datanomics model to exponentially decline in cost and performance.
Figure 1: Physical Business Datanomics
Datanomics and the digital enterprise
Data is at the heart of a digital enterprise, and there is a constant iterative process of data creation, processing, analysis and sharing/selling. Data created by an IoT sensor or social media app has little value at the time of creation. The value increases exponentially when it is then used to develop applications, further analyzed for targeted outcomes, and sold to consumers or other vendors. Even the database in an Oracle Exadata appliance becomes more business critical as hundreds of developers and data scientists need to access and analyze the data.
In a digital enterprise, data is constantly accessed by multiple applications throughout its lifecycle, and this completely upends the Datanomics model compared to a physical enterprise. The impact on the value of data is transformative, as the chart below shows.
Pharma and healthcare companies leverage data for dramatically more efficient drug discovery and patient management process. And some of the most valuable data in drug discovery is historical. Data about user environment, behavior, market, and competitive landscape is not only critical for the market leadership of Netflix, but for its very survival.
Data value does not conveniently fit into an exponential-decay-with-time model anymore and is much more dynamic based on various applications using data throughout its lifecycle. Access needs to be consistently fast for use by application development, analytics, publishing, etc. while meeting the needs of an enterprise to be inherently protected, always resilient and compliant throughout its lifecycle.
Figure 2: Digital Enterprise Datanomics
Managing Datanomics of digital transformation
Enterprises undertaking digital transformation have a significant challenge of simultaneously managing traditional predictable systems and applications while enabling rapid application development and data delivery for innovation and experimentation. Gartner calls this Bimodal IT, and Marc Andreessen summarized how this transformation is happening in his seminal article “Software is Eating the World.”
In a physical business, application development and IT operations were distinct processes, with a formal request-response engagement between the two. Application developers requested infrastructure or data, and the operations team provisioned servers, storage, and networks and configured proprietary software to provision and manage data. The entire process took several days, which when combined with the “waterfall” software development model, resulted in long cycle times. A physical-centric IT organization addressed the needs of a physical-centric business.
Digital enterprises require continuous development of core business applications and continuous management, delivery and analysis of data. Data needs to be immediately available for app testing and analytics if the business is to successfully evolve and grow. A DevOps IT model with a continuous integration and continuous deployment (CI/CD) software development model and use of analytics or machine learning tools to analyze data are an integral part of a digital enterprise.
Most organizations undertaking digital transformation have responded by addressing two components of the stack:
- Virtualize infrastructure, delivering infrastructure as a service (IaaS) and provision infrastructure on demand via APIs.
- Virtualize the application software to deliver software as a service (SaaS).
When it comes to the third, key component of the stack, the only proprietary asset of the enterprise, data, enterprises respond by applying the tools, processes and products of a traditional physical business to meet the requirements of a digital enterprise. And they fail miserably.
Figure 3: Traditional tools attempting to deliver digital transformation
Data is manually extracted at different stages of the lifecycle for use by application development, analytics, sharing, selling, compliance or other use cases. Extracting and transforming data each time is limited by the data “gravity” problem. It requires significant time and resources every time and, depending on the size of data, could take days, weeks or months.
Instant infrastructure availability without instant data availability results in a zombie digital enterprise with all the skeleton and no life.
Digital transformation has also created a secular disruption of the storage industry by breaking the foundational model it was built on. By using the traditional data management tools to deliver instant access to data anywhere, from any point in time, requires the fastest and the most expensive storage all the time, everywhere! It’s an unsustainable economic model.
For almost a decade now, industry vendors have invested billions in building newer storage products and business models: high-performance, scale-out, all-flash, hybrid, cloud, software-defined—the list goes on. They fundamentally overlooked the fact that digital transformation also happened to the storage industry. It has nothing to do with storage. And everything to do with data.
Datanomics for the digital enterprise requires a different data management model, one that is virtualized and combines the needs of the operations teams, who are custodians of data, and the developers and data scientists, who are the consumers of data.
In the next blog post, I’ll take a deeper look at the new data model to deliver on the requirements of Datanomics of modern, digital enterprises. We will look at specific examples of enterprises that have implemented some of the organizational changes needed to enable this model.
Read the article on Infoworld