Latency-sensitive applications such as Artificial Intelligence, Machine Learning, and High-Frequency Trading create an insatiable need for performance. While the performance promises of NVMe are real, legacy storage architectures choke NVMe performance at the controller or force inefficient and uneconomical workarounds such as Software-Defined Storage (SDS) or Direct Attached Storage (DAS). This no longer needs to be the case.
Industry experts have predicted the inevitability of Composable Disaggregated Infrastructure as it becomes clear that separating compute, storage and networking resources into shared resource pools accessed with low-latency protocols like NVMe-oF deliver the performance and economics needed to support the scale of modern application growth. Read this white paper to learn more about disaggregating storage from compute in scale-out environments.
