The unstoppable data growth trend of our era is further fueled by cutting-edge technologies such as Image Recognition, IoT & Edge Computing, Virtual / Augmented Reality, Neural Networks and Big Data Analytics.
Those novel use cases generate, ingest and compute tremendous amounts of data. Whether it is real-time or background processing, the traditional mechanisms to do so are no longer suitable or optimal enough to satisfy the requirements of these applications: high throughput, high bandwidth and ultra-low latency response times (near real-time). They require large amounts of compute in environments where miniaturization and footprint reduction are often critical, and where bandwidth may be the contention factor.
Computational storage is the storage industry response to address these unique challenges. With data becoming one of the most precious resources, and with the mainstream adoption of these use cases in the next five years, organizations should get acquainted with the basic principles of computational storage.
In this industry research document, TECHunplugged covers the challenges faced by traditional storage when exposed to these emerging use cases. We will also cover the solutions brought forth by computational storage, an overview of differences between this novel approach and traditional storage, to end with an analysis of the current market landscape.