Search Results

Mixture of Expert (MoE) models consist of several experts, each specializing in a specific task. During inference, a subset of the experts is invoked based on their relevance to the request. MoE's modular architecture lets users compose their model from ...

Deep learning (DL) training is data-intensive and often bottlenecked by fetching data from remote storage. Recognizing that many samples' sizes diminish during data preprocessing, we explore selectively offloading preprocessing to remote storage to ...

Active-standby schemes for Virtual Machine (VM) high availability require periodic synchronization of memory and CPU state. The most common approach to synchronization is to use page tables and software to identify "dirty" memory pages at the source and ...

Fail-slow hardware is an under-studied failure mode. We present a study of 114 reports of fail-slow hardware incidents, collected from large-scale cluster deployments in 14 institutions. We show that all hardware types such as disk, SSD, CPU, memory, ...

As the influence of machine learning grows over decisions in businesses and human life, so grows the need for Model Governance. In this paper, we motivate the need for, define the problem of, and propose a solution for Model Governance in production ML. ...

FEMU is a software (QEMU-based) flash emulator for fostering future full-stack software/ hardware SSD research. FEMU is cheap (opensourced), relatively accurate (0.5-38% variance as a drop-in replacement of OpenChannel SSD), scalable (can support 32 ...

Fail-slow hardware is an under-studied failure mode. We present a study of 101 reports of fail-slow hardware incidents, collected from large-scale cluster deployments in 12 institutions. We show that all hardware types such as disk, SSD, CPU, memory and ...

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 instability is the well-known garbage collection (GC) process, which ...

TTFLASH is a "tiny-tail" flash drive (SSD) that eliminates GC-induced tail latencies by circumventing GC-blocked I/Os with four novel strategies: plane-blocking GC, rotating GC, GC-tolerant read, and GC-tolerant flush. It is built on three SSD internal ...

Host-side flash caching has emerged as a promising solution to the scalability problem of virtual machine (VM) storage in cloud computing systems, but it still faces serious limitations in capacity and endurance. This paper presents CloudCache, an on-...

As flash devices become ubiquitous in data centers and cost per gigabyte drops, flash systems need to provide data services similar to those of traditional storage. We present Mjölnir, a powerful and scalable engine that addresses the core problems that ...

The emergence of persistent memories promises a sea-change in application and data center architectures, with efficiencies and performance not possible with today's volatile DRAM and persistent slow storage. We present Software Defined Persistent Memory,...

Key-value stores are ubiquitous in high performance data-intensive, scale out, and NoSQL environments. Many KV stores use flash devices for meeting their performance needs. However, by using flash as a simple block device, these KV stores are unable to ...

We present a new form of storage virtualization based on block-level address remapping. By allowing the host system to manipulate this address map with a set of three simple operations (clone, move, and delete), we enable a variety of useful features ...

State-of-the-art flash-optimized KV stores frequently rely upon a log structure and/or compaction-based strategy to optimally organize content on flash. However, these strategies lead to excessive I/O, beyond the write amplification generated within the ...

Snapshots are a common and heavily relied upon feature in storage systems. The high performance of flash-based storage systems brings new, more stringent, requirements for this classic capability. We present ioSnap, a flash optimized snapshot system. ...

Flash memory is widely used for its fast random I/O access performance in a gamut of enterprise storage applications. However, due to the limited endurance and asymmetric write performance of flash memory, minimizing writes to a flash device is critical ...

Host-side flash-based caching offers a promising new direction for optimizing access to networked storage. Current work has argued for using host-side flash primarily as a read cache and employing a write-through policy which provides the strictest ...

We present anticipatory memory allocation (AMA), a new method to build kernel code that is robust to memory-allocation failures. AMA avoids the usual difficulties in handling allocation failures through a novel combination of static and dynamic ...