<< back Page 3 of 3

Data Moves Closer to the Edge, And it's a Good Thing


PLANNING AHEAD FOR IOT AND EDGE

What will an edgy future look like? Enterprises need to plan ahead for an explosion on edge and IoT computing, and this starts with an architecture that is tiered and supports both structured and unstructured data types. “Data will be located and acted on locally, at the edge, as close to the data source as possible,” Currie predicted. “Data will also need to be aggregated and stored regionally or centrally. The centralized data will look at macro patterns and interact with other enterprise data sources for greater insights and action. Using the manufacturing example, IoT data will affect the local processing, but then the aggregated data may be used to manage supplier relationships.”

While much of edge data is currently comprised of time-series and images, videos, or IoT data, future data structures will also need to support “other semi-structured or unstructured data—such as device metadata and audio,” said Jain. “So, in reality, most enterprises will have to support heterogeneous data structures and analysis.”

Meeting the requirements for an edge driven data structure is “more about making sure that data fusion can happen as quickly as possible to support decision making,” said Lewis. “This translates into minimization of supported data formats along with translators between data formats where necessary, serializable data types that lend themselves to data streaming, and use of simple, rather than complex, encodings where possible, like, for example, choosing JSON over XML when both are feasible options. In addition, aligning IoT data formats to the data schemas used by the database where this data is stored will also simplify the storage, retrieval, and analysis of IoT data.”

Many organizations will likely move to hybrid solutions that support cloud-based, on-premise deployments, and edge. “A hybrid solution with some sort of local storage can provide a valuable degree of control that complements the flexibility of the cloud,” said Nilsson. “Local storage solutions mean users have a greater degree of control over image quality and retention.”

<< back Page 3 of 3


Newsletters

Subscribe to Big Data Quarterly E-Edition