Technology Vision 2023 | When Atoms meet Bits #TechVision down approach that installs a virtual layer on Data mesh is another emerging concept which top of an organization’s various data sources for tackles the problem in a different way. If data fabrics tightly controlled, uni昀椀ed, data management. try to assert top-down management over data, data Metaphorically speaking, it is like building a mesh treats each data source as an independent zoo around your data. You aren’t changing the product. Coined by Zhamak Dehghani in 2019, underlying data, and in fact you’re trying to preserve data mesh stresses domain-speci昀椀c expertise, 145 Data mesh is another the environments—but at the same time you’re management, and governance. The owners of data building consistency across the enclosures, better sources are distributed and entrusted to manage emerging concept which descriptions about what’s inside, and a map to where data themselves, with the expectation that they tackles the problem everything is so it’s all readily accessible by everyone. are trying to reduce the friction to access that data in a di†erent way. (typically through the use of APIs). The major boon here is how the single source of truth and automation can dramatically increase This is particularly useful for businesses that have productivity. One survey found data scientists varied data needs across different environments. spend 45% of their time on data preparation tasks, For instance, an oil and gas company may seek to 143 including data loading and data cleansing. While understand offshore extraction data at one site, these steps may be time-consuming and tedious, then pipeline ef昀椀ciency data at another. So, instead they are necessary to ensure data quality for things of something like a data lake pooling information like data visualization or model development. A data together, a data mesh paradigm can help a company a solid data management strategy is necessary in fabric paradigm can automate data tasks, saving access data across locations from multiple sources. this new era of transparency. That said, the ability the enterprise and its employees’ valuable time, to unlock “windows of transparency” is not the thanks to an active metadata management system. No matter which data strategy you end up with (and same thing as actually opening them. Indeed, many Gartner predicts that by 2025, “active metadata- it can even be both), data mesh and data fabric, businesses will choose to keep their insights under assisted automated functions in the data fabric will as well as the technologies behind them, can help lock-and-key—but as exempli昀椀ed by Yelp, they may reduce human effort by a third, while improving data your enterprise manage its data better—and having 昀椀nd their data out in the open anyway, and soon. 144 utilization fourfold.”