SK hynix announced on November 16 (local time) that its technical team won the IT/Electrical category at the 2023 R&D 100 Awards held in San Diego, California, USA, for developing the Key Value Computational Storage Drive (KV-CSD), a next-generation storage device.
The R&D 100 Awards is a science and technology award ceremony that annually recognizes the 100 most innovative technologies and products in the world, and is known as the "Oscars of Innovation" in the industry.
KV-CSD is a product developed by SK hynix in collaboration with Los Alamos National Laboratory (LANL) in the U.S. SK hynix developed KV-CSD and LANL developed a system to apply KV-CSD to supercomputers.
The product shows fast read-write performance that can process large-scale data of several petabytes (PB) in minutes, and can quickly analyze data with its own computation, which has attracted great attention in the field of high-performance computing (HPC).
SK hynix also emphasized that it has applied the world's first "Indexing" technology to KV-CSD, which dramatically speeds up data retrieval.
Normally, when saving a file, the data is divided and stored in multiple storage spaces, called blocks, but by utilizing the indexing technology, the data that is divided into multiple blocks is given a unique key, and if necessary, it can be quickly collected and processed according to the indexing criteria.
The speed of KV-CSD is up to 7.4 times faster than solid state drives (SSDs), which are conventional storage devices that collect and process required data by looking at blocks one by one, the company said.
"KV-CSD is a next-generation storage device that quickly processes data based on key values and indexing, analyzes it, and delivers only the desired results to the CPU, which will be useful for high-performance computing as well as the AI and big data industries that process large amounts of data," said Woo-seok Jung, head of the Computational Storage Team at SK hynix.
Several concepts are currently being developed for next-generation storage. They are evolving in various forms, including those that process data close to the CPU, or those that compute data directly instead of the CPU. Storage devices that directly compute data are called computational storage drives (CSDs), and SK hynix's KV-CSD falls into this category.
The company emphasized that KV-CSD is also significant in that it solves customer pain points. SK hynix added that the product is an example of a solution provider that dramatically reduces the time required to process big data.
