This new open-source version incorporates flexible configurations that allow system architects to fine-tune the balance between search performance and the number of vectors—two factors that compete for space within a fixed SSD capacity. The resulting benefit allows RAG system designers to precisely adjust the optimal balance for specific workloads and their requirements without requiring hardware modifications.
The KIOXIA AiSAQ software, first introduced in January 2025, uses a novel Approximate Nearest Neighbor Search (ANNS) algorithm optimized for SSDs, eliminating the need to store index data in DRAM. By enabling vector searches directly on SSDs and reducing host memory requirements, KIOXIA AiSAQ technology allows vector databases to scale without being limited by restricted DRAM capacity.
When the installed SSD capacity in the system is fixed, increasing search performance (queries per second) requires higher SSD capacity consumption per vector. The result is a smaller number of vectors. Conversely, maximizing the number of vectors requires reducing SSD capacity consumption per vector, which leads to lower performance. The optimal balance between these two opposing conditions varies depending on the specific workload. To find the right balance, KIOXIA AiSAQ software introduces flexible configuration options. This latest update allows administrators to select the optimal balance point for a wide variety of contrasting workloads within the RAG system. This update transforms KIOXIA AiSAQ into an SSD-based ANNS solution that is not only ideal for RAG environments but also for other vector-intensive applications, such as offline semantic searches.
Given the growing demand for scalable AI solutions, SSDs are emerging as an efficient alternative to DRAM to meet the high-speed, low-latency requirements of RAG systems. KIOXIA AiSAQ software makes it possible to efficiently meet these demands by enabling large-scale generative AI without being limited by memory resources.
By releasing the KIOXIA AiSAQ software as open source, KIOXIA reinforces its commitment to the artificial intelligence community by promoting SSD-centric architectures for scalable AI.
“We are always looking for innovative ways to help developers and system architects optimize performance and capacity,” commented Axel Störmann, Vice President and Chief Technology Officer for Memory and SSD products at KIOXIA Europe GmbH. “Now, with this new version of the AiSAQ software, users can fully leverage the potential of SSDs to design scalable, flexible, and efficient RAG systems. By releasing this technology as open source, we are reaffirming our unwavering commitment to the AI community by providing them with powerful and accessible solutions.”
Follow the link to download the open-source KIOXIA AiSAQ software.
