Efficiently scheduling data processing jobs on distributed compute clusters requires complex
algorithms. Current systems use simple, generalized heuristics and ignore workload …

We introduce Gandiva, a new cluster scheduling framework that utilizes domain-specific
knowledge to improve latency and efficiency of training deep learning models in a GPU …

Existing deep learning systems commonly parallelize deep neural network (DNN) training
using data or model parallelism, but these strategies often result in suboptimal …

Scheduling plays a pivotal role in cloud computing systems. Designing an efficient
scheduler is a challenging task. The challenge comes from several aspects, including the …

Deep learning (DL) training jobs bring some unique challenges to existing cluster
managers, such as unpredictable training times, an all-or-nothing execution model, and …

Deep learning workloads are common in today's production clusters due to the proliferation
of deep learning driven AI services (eg, speech recognition, machine translation). A deep …

Operating data-intensive applications on edge systems is challenging, due to the extreme
workload and device heterogeneity, as well as the geographic dispersion of compute and …

Machine learning inference is becoming a core building block for interactive web
applications. As a result, the underlying model serving systems on which these applications …

Serverless computing is becoming increasingly popular due to its ease of programming, fast
elasticity, and fine-grained billing. However, the serverless provider still needs to provision …

We describe the design and implementation of Protean--the Microsoft Azure service
responsible for allocating Virtual Machines (VMs) to millions of servers around the globe. A …