Palacios and Kitten: New high performance operating systems for scalable virtualized and native supercomputing

Lange, John R.; Pedretti, Kevin; Hudson, Trammell; Dinda, Peter A.; Cui, Zheng; Lei, Xia; Bridges, Patrick G.; Gocke, Andy; Jaconette, Steven; Levenhagen, Mike; Brightwell, Ron

doi:10.1109/ipdps.2010.5470482

Cited by 126 publications

(78 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kitten is similar to previous lightweight kernels (LWK) such as SUNMOS, Puma, Cougar, and Catamount. Kitten, however, distinguishes itself from these prior LWKs by providing a Linux-compatible user environment, a more modern and extensible code base, and a virtual machine monitor capability via the Palacios virtual machine monitor [30] which allows full-featured guest operating systems to be loaded on-demand at very low overhead [26].…”

Section: Approachmentioning

confidence: 99%

Evaluating operating system vulnerability to memory errors

Ferreira

Pedretti

Brightwell

et al. 2012

Proceedings of the 2nd International Workshop on Runtime and Operating Systems for Supercomputers

Self Cite

View full text Add to dashboard Cite

Section: Approachmentioning

confidence: 99%

Evaluating operating system vulnerability to memory errors

Ferreira

Pedretti

Brightwell

et al. 2012

Proceedings of the 2nd International Workshop on Runtime and Operating Systems for Supercomputers

Self Cite

View full text Add to dashboard Cite

“…Specifically, custom OSs have thrived in high performance computing (HPC) [34,40]. For example, Kitten [31] is a lightweight kernel that typically runs on the compute nodes of supercomputers. Kitten focuses on reducing kernel interference and maximizing performance.…”

Section: Related Workmentioning

confidence: 99%

Improving per-node efficiency in the datacenter with new OS abstractions

Rhoden

Klues

Zhu

et al. 2011

Proceedings of the 2nd ACM Symposium on Cloud Computing

View full text Add to dashboard Cite

We believe datacenters can benefit from more focus on per-node efficiency, performance, and predictability, versus the more common focus so far on scalability to a large number of nodes. Improving per-node efficiency decreases costs and fault recovery because fewer nodes are required for the same amount of work. We believe that the use of complex, general-purpose operating systems is a key contributing factor to these inefficiencies.Traditional operating system abstractions are ill-suited for high performance and parallel applications, especially on large-scale SMP and many-core architectures. We propose four key ideas that help to overcome these limitations. These ideas are built on a philosophy of exposing as much information to applications as possible and giving them the tools necessary to take advantage of that information to run more efficiently. In short, high-performance applications need to be able to peer through layers of virtualization in the software stack to optimize their behavior. We explore abstractions based on these ideas and discuss how we build them in the context of a new operating system called Akaros.

show abstract

“…Virtualization has the potential to dramatically increase the usability and reliability of high performance computing (HPC) systems by maximizing system flexibility and utility to a wide range of users [16,21,18]. We argue here that the virtualization infrastructure of a large-scale HPC system should include a facility that continuously tracks memory content sharing across the machine.…”

Section: Introductionmentioning

confidence: 99%

A case for tracking and exploiting inter-node and intra-node memory content sharing in virtualized large-scale parallel systems

Lei

Dinda

2012

Proceedings of the 6th International Workshop on Virtualization Technologies in Distributed Computing Date

Self Cite

View full text Add to dashboard Cite

In virtualized large-scale parallel systems scientific workloads consist of numerous processes running across many virtual nodes. Their memory footprint is massive, and this has consequences for services that enhance performance, reliability, or power. We argue that a service that dynamically tracks the sharing of memory content, both within individual nodes, and across nodes, can simplify and enhance the implementation of such services. For example, leveraging content sharing could significantly reduce the size of a checkpoint of a group of nodes. As another example, it could speed VM migration by allowing the reconstruction of a VM's memory from multiple source VMs. Finally, a service that improves reliability by introducing memory redundancy could leverage existing content sharing to minimize the memory costs of any particular level of redundancy. We argue that both intra-and inter-node memory content sharing is common in parallel applications, supporting this claim by a detailed study of both kinds of sharing, at different scales, different granularities, and different times for a range of applications and application benchmarks. We then describe the high level approach we are taking to design and implement a distributed, VMM-based system that can efficiently and scalably identify and track such sharing with low overhead.

show abstract

Palacios and Kitten: New high performance operating systems for scalable virtualized and native supercomputing

Cited by 126 publications

References 18 publications

Evaluating operating system vulnerability to memory errors

Evaluating operating system vulnerability to memory errors

Improving per-node efficiency in the datacenter with new OS abstractions

A case for tracking and exploiting inter-node and intra-node memory content sharing in virtualized large-scale parallel systems

Contact Info

Product

Resources

About