Research Summary: 1 Introduction Networked systems are ubiquitous in our daily lives, playing a critical role across a wide range of scientific fields, including communication, machine learning, optimization, control, biology, economics, and social ...

This work highlights the significance of I/O bottlenecks that data-intensive HPC workflows face in serverless environments - an issue that has been largely overlooked by prior works. We propose StarShip, a framework that leverages different storage ...

Processing-in-memory (PIM) has been widely explored in academia and industry to accelerate numerous workloads. By reducing the data movement and increasing parallelism, PIM offers great performance and energy efficiency. A large amount of cores or nodes ...

This study demonstrates the salient facts and challenges of host failure operations in hyperscale data centers. A host incident can involve hundreds of distinct host-level metrics. The faulting mechanism inside the host connects these heterogeneous ...

GPUs execute thousands of active threads simultaneously, requiring high memory bandwidth to handle multiple memory requests efficiently. The memory bandwidth in GPUs has always been increasing, but it is still insufficient for the demands of fine-grained ...

Buffer overflow is a widespread memory safety violation in C/C++, reported as the top vulnerability in 2022. Secure memory allocators are generally used to protect systems against attacks that may exploit buffer overflows. Existing allocators mainly rely ...

Due to inherent variations in energy's carbon intensity, temporal shifting has become a key method in reducing the carbon footprint of batch workloads. However, temporally shifting workloads involves searching for periods with lower carbon intensity, ...

As robots increasingly permeate modern society, it is crucial for the system and hardware research community to bridge its long-standing gap with robotics. This divide has persisted due to the lack of (i) a systematic performance evaluation of robotics ...

The InterPlanetary File System (IPFS) has recently gained considerable attention. While prior research has focused on understanding its performance characterization and application support, it remains unclear: (1) what kind of files/content are stored in ...

Serverless (or Function-as-a-Service) compute model enables new applications with dynamic scaling. However, all current Serverless systems are best-effort, and as we prove this means they cannot guarantee hard real-time deadlines, rendering them ...

Resource management in microservices is challenging due to the uncertain latency-resource relationship, dynamic environment, and strict Service-Level Agreement (SLA) guarantees. This paper presents a Pessimistic and Optimistic Bayesian Optimization ...

Abstract-Multi-modal deep neural networks (DNNs) have become increasingly pervasive in many machine learning application domains due to their superior accuracy by fusing various modalities together. However, multi-modal DNNs present many unique ...

Motivation. Microservices architecture has become more prevalent in cloud-based applications where small, looselycoupled service components work together to handle user requests [1]. This architecture differs greatly from traditional monolithic ...

The growing adoption of hardware accelerators driven by their intelligent compiler and runtime system counterparts has democratized ML services and precipitously reduced their execution times. This motivates us to shift our attention to characterize the ...

Cloud computing represents an appealing opportunity for cost-effective deployment of HPC workloads on the best-fitting hardware. However, although cloud and on-premise HPC systems offer similar computational resources, their network architecture and ...

We present Memtrade, the first practical marketplace for disaggregated memory clouds. Clouds introduce a set of unique challenges for resource disaggregation across different tenants, including resource harvesting, isolation, and matching. Memtrade ...

We consider the problem of balancing the load among servers in dense racks for microsecond-scale workloads. To balance the load in such settings, tens of millions of scheduling decisions have to be made per second. Achieving this throughput while ...

Training deep learning (DL) models in the cloud has become a norm. With the emergence of serverless computing and its benefits of true pay-as-you-go pricing and scalability, systems researchers have recently started to provide support for serverless-...

Modern cloud block service provides cloud users with virtual block disks (VDisks), and it usually relies on a forwarding layer consisting of multiple proxy servers to forward the block-level writes from applications to the underlying distributed storage. ...

Traditional data centers include monolithic servers that tightly integrate CPU, memory and disk (Figure 1a). Instead, Disaggregated Systems (DSs) [8, 13, 18, 27] organize multiple compute (CC), memory (MC) and storage devices as independent, failure-...