The judgment of forseti: economic utility for dynamic heap sizing of multiple runtimes

Cameron, Callum; Singer, Jeremy; Vengerov, David

doi:10.1145/2754169.2754180

Cited by 11 publications

(3 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Efforts to efficient management of independent Java VM heaps led to dynamic allocation of resources shared between the running virtual machines [13,14,45]. These studies differ from our study in that (i) our system focuses on a restricted mobile platform that hosts dozens of parallel virtual machines, (ii) we evaluate the GC as perceived globally on the system profiler, and (iii) the GC service performs GC phases on all virtual machines.…”

Section: Related Workmentioning

confidence: 95%

One Process to Reap Them All

et al. 2017

View full text Add to dashboard Cite

Ubiquitous mobile platforms such as Android rely on managed language run-time environments, also known as language virtual machines (VMs), to run a diverse range of user applications (apps). Each app runs in its own private VM instance, and each VM makes its own private local decisions in managing its use of processor and memory resources. Moreover, the operating system and the hardware do not communicate their low-level decisions regarding power management with the high-level app environment. This lack of coordination across layers and across apps restricts more effective global use of resources on the device. We address this problem by devising and implementing a global memory manager service for Android that optimizes memory usage, run-time performance, and power consumption globally across all apps running on the device. The service focuses on the impact of garbage collection (GC) along these dimensions, since GC poses a significant overhead within managed run-time environments. Our prototype collects system-wide statistics from all running VMs, makes centralized decisions about memory management across apps and across software layers, and also collects garbage centrally. Furthermore, the global memory manager coordinates with the power manager to tune collector scheduling. In our evaluation, we illustrate the impact of such a central memory management service in reducing total energy consumption (up to 18%) and increasing throughput (up to 12%), and improving memory utilization and adaptability to user activities.

show abstract

Section: Related Workmentioning

confidence: 95%

One Process to Reap Them All

et al. 2017

View full text Add to dashboard Cite

show abstract

“…However, these JVMs are targeted at monolithic applications, not distributed workloads. There are also projects that share some of our goals: Forseti [24] investigates holistic heap sizing. The MVM [41] looked at running multiple applications in the same JVM.…”

Section: Related Workmentioning

confidence: 99%

Taurus

Maas

Asanović

Harris

et al. 2016

SIGARCH Comput. Archit. News

View full text Add to dashboard Cite

Many distributed workloads in today's data centers are written in managed languages such as Java or Ruby. Examples include big data frameworks such as Hadoop, data stores such as Cassandra or applications such as the SOLR search engine. These workloads typically run across many independent language runtime systems on different nodes.This setup represents a source of inefficiency, as these language runtime systems are unaware of each other. For example, they may perform Garbage Collection at times that are locally reasonable but not in a distributed setting.We address these problems by introducing the concept of a Holistic Runtime System that makes runtime-level decisions for the entire distributed application rather than locally. We then present Taurus, a Holistic Runtime System prototype. Taurus is a JVM drop-in replacement, requires almost no configuration and can run unmodified off-the-shelf Java applications. Taurus enforces user-defined coordination policies and provides a DSL for writing these policies.By applying Taurus to Garbage Collection, we demonstrate the potential of such a system and use it to explore coordination strategies for the runtime systems of real-world distributed applications, to improve application performance and address tail-latencies in latency-sensitive workloads.

show abstract

“…In the original presentation of agoric systems, the authors focus on specialized distributed computation activities, such as garbage collection across trust boundaries [22]. We have also developed simple economic models for memory management in virtual machines [23], [24]. However broadly similar principles should be applicable to the more general resource management issues in IaaS cloud provision.…”

Section: Related Workmentioning

confidence: 99%

MyHealthAvatar: A Case Study of Web-Based Interactive Visual Analytics of Lifestyle Data

Parvinzamir

Zhao

Deng

et al. 2015

2015 IEEE International Conference on Computer and Information Technology; Ubiquitous Computing and Communications; Dependable,

View full text Add to dashboard Cite

Although cloud computing has made great inroads in terms of consumer adoption, we argue that the paradigm still has a long way to go to reach an economically efficient free market state. We highlight four key problems with the current cloud computing approach. These are (1) coarse granularity of service provision, (2) limited control of resources, (3) inadequate support for small providers and prosumers, and (4) vendor lockin via APIs. In this position paper, we briefly outline interesting work that may lead to a solution, in terms of moving towards a more market-based approach to cloud resource allocation.

show abstract

The judgment of forseti: economic utility for dynamic heap sizing of multiple runtimes

Cited by 11 publications

References 22 publications

One Process to Reap Them All

One Process to Reap Them All

Taurus

MyHealthAvatar: A Case Study of Web-Based Interactive Visual Analytics of Lifestyle Data

Contact Info

Product

Resources

About