We present WiscKey, a persistent LSM-tree-based key-value store with a performance-
oriented data layout that separates keys from values to minimize I/O amplification. The …

Flash storage has become the mainstream destination for storage users. However, SSDs do
not always deliver the performance that users expect. The core culprit of flash performance …

The conventional wisdom in systems and networking communities is that congestion
happens primarily within the network fabric. However, adoption of high-bandwidth access …

This paper presents LinnOS, an operating system that leverages a light neural network for
inferring SSD performance at a very fine—per-IO—granularity and helps parallel storage …

We study storage performance in over 450,000 disks and 4,000 SSDs over 87 days for an
overall total of 857 million (disk) and 7 million (SSD) drive hours. We find that storage …

Key-value store based on a log-structured merge-tree (LSM-tree) is preferable to hash-
based KV store because an LSM-tree can support a wider variety of operations and show …

Predictable latency on flash storage is a long-pursuit goal, yet, unpredictability stays due to
the unavoidable disturbance from many well-known SSD internal activities. To combat this …

We present SWAN, a novel All Flash Array (AFA) management scheme. Recent flash SSDs
provide high I/O bandwidth (eg, 3-10GB/s) so the storage bandwidth can easily surpass the …

Graph neural networks (GNNs) process large-scale graphs consisting of a hundred billion
edges. In contrast to traditional deep learning, unique behaviors of the emerging GNNs are …

On 20 March, 1974, Professor Edward Banfield, of the University of Pennsylvania, was
prevented from delivering a lecture at the University of Chicago. Professor Banfield, who is a …