Quasar: resource-efficient and QoS-aware cluster management

By Christina Delimitrou, Christos Kozyrakis. International Conference on Architectural Support for Programming Languages and Operating Systems, 2014

1,134 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Hardware and Architecture, Information Systems

View Paper: https://www.semanticscholar.org/paper/3000e77ed7282d9fb27216f3e862a3769119d89e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Hardware and Architecture

In Information Systems

Citations in the corpus

Ernest: Efficient Performance Prediction for Large-Scale Advanced Analytics (2016, methodology): “Finally scheduling frameworks like Quasar [36] try to estimate the scale out and scale up factor for jobs using the progress rate of the ﬁrst few tasks.”
Planaria: Dynamic Architecture Fission for Spatial Multi-Tenant Acceleration of Deep Neural Networks (2020, background): “There is a large swath of related work on multi-tenancy for CPUs [36, 66, 97–103] and GPUs [29, 35, 37–42, 66, 104–110] due to its vitality for cloud-scale computing.”
Reinforcement Learning for Primary care e Appointment Scheduling (2017, background): “The system is complex and often impossible to model accurately [2], [3]; 2.”
Domain-specific Genetic Algorithm for Multi-tenant DNNAccelerator Scheduling (2021, background): “The CPU host packs independent jobs (layers) into a ready-batch and queries the optimizer for schedule, making it a static scheduling problem, whose procedure is consistent with the assumption in many resource scheduling works [14, 23, 66].”
Flexible Scheduling of Distributed Analytic Applications (2016, methodology): “This can be easily included in an application description, but requires developing additional components to infer statistical models of application properties based on systems observations, as done for example in Tarcil [19], Paragon [15] and Quasar [16].”
Resource Management with Deep Reinforcement Learning (2016, background): “3005750 server characteristics, interactions with other tasks, and interference on shared resources such as CPU caches, network bandwidth, etc [12, 17].”
Resource Central: Understanding and Predicting Workloads for Improved Resource Management in Large Cloud Platforms (2017, background): “Many of these works use online predictions of resource usage or performance interference [3, 7, 20, 27].”
Who Limits the Resource Efficiency of My Datacenter: An Analysis of Alibaba Datacenter Traces (2019, background): “In order to guarantee the QoS of LC applications, a common approach is reserving ample resources for them [7].”
CherryPick: Adaptively Unearthing the Best Cloud Configurations for Big Data Analytics (2017, background): “Online scheduler of applications: Paragon [15] and Quasar [16] are online schedulers that leverage historical performance data from scheduled applications to quickly classify any new incoming application, assign the application proper resources in a datacenter, and reduce interferences among differe”
Seer: Leveraging Big Data to Navigate the Complexity of Performance Debugging in Cloud Microservices (2019, background): “Again consistent with queueing theory [49] and prior work [34, 37, 38, 46, 56], per-microservice queue depths accurately capture performance bottlenecks and pinpoint the Session: Cloud I ASPLOS’19, April 13–17, 2019, Providence, RI, USA”
FIRM: An Intelligent Fine-Grained Resource Management Framework for SLO-Oriented Microservices (2020, background): “Examples include Borg [119], Mesos [43], Tarcil [28], Paragon [25], Quasar [26], Morpheus [54], DeepDive [73], and Q-clouds [71].”
Heterogeneity-Aware Cluster Scheduling Policies for Deep Learning Workloads (2020, methodology): “Gavel makes these scheduling decisions transparently: it specifies an API between the scheduler and applications that allow jobs written in existing deep learning frameworks like PyTorch [48] and TensorFlow [13] to be moved between resources with minimal code changes, and uses a mechanism similar to”
Heracles: Improving resource efficiency at scale (2015, methodology): “Batch analytics frameworks can generate numerous BE tasks and derive significant value even if these tasks are occasionally deferred or restarted [19, 10, 13, 16].”
Pocket: Elastic Ephemeral Storage for Serverless Analytics (2018, background): “The result is significant capacity underutilization at the cluster level [21, 74, 29].”