iBalloon: Efficient VM Memory Balancing as a Service

Zhang, Qi; Liu, Ling; Ren, Jiangchun; Su, Gong; Iyengar, Arun

doi:10.1109/icws.2016.14

Cited by 10 publications

(4 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Policies for dynamic VM memory sizing face the fundamental size vs performance tradeoff, using techniques such as working set estimation [8,60], and application and OS performance metrics [16,19,50,56]. Our work focuses on memory harvesting in production public cloud platforms, where such techniques are considered invasive and cannot be universally and safely applied.…”

Section: Hypervisor-level Memory Managementmentioning

confidence: 99%

Memory-harvesting VMs in cloud platforms

Fuerst

Novaković

Goiri

et al. 2022

Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems

View full text Add to dashboard Cite

Cloud platforms monetize their spare capacity by renting "Spot" virtual machines (VMs) that can be evicted in favor of higher-priority VMs. Recent work has shown that resource-harvesting VMs are more effective at exploiting spare capacity than Spot VMs, while also reducing the number of evictions. However, the prior work focused on harvesting CPU cores while keeping memory size fixed. This wastes a substantial monetization opportunity and may even limit the ability of harvesting VMs to leverage spare cores. Thus, in this paper, we explore memory harvesting and its challenges in real cloud platforms, namely its impact on VM creation time, NUMA spanning, and page fragmentation. We start by characterizing the amount and dynamics of the spare memory in Azure. We then design and implement memory-harvesting VMs (MHVMs), introducing new techniques for memory buffering, batching, and pre-reclamation. To demonstrate the use of MHVMs, we also extend a popular cluster scheduling framework (Hadoop) and a FaaS platform to adapt to them. Our main results show that (1) there is plenty of scope for memory harvesting in real platforms; (2) MHVMs are effective at mitigating the negative impacts of harvesting; and (3) our extensions of Hadoop and FaaS successfully hide the MHVMs' varying memory size from the users' data-processing jobs and functions. We conclude that memory harvesting has great potential for practical deployment and users can save up to 93% of their costs when running workloads on MHVMs. CCS CONCEPTS• Computer systems organization → Cloud computing.

show abstract

Section: Hypervisor-level Memory Managementmentioning

confidence: 99%

Memory-harvesting VMs in cloud platforms

Fuerst

Novaković

Goiri

et al. 2022

Proceedings of the 27th ACM International Conference on Architectural Support for Programming Languages and Operating Systems

View full text Add to dashboard Cite

show abstract

“…Many approaches for performance-sensitive resource allocation among co-located VMs have been suggested [22,24,28,32,55], but they assume some application performance model, which our work does not. VM memory allocations can be set using working-set estimation [13,53,54], utility-maximizing [25], or market-based approaches [6,11]. As noted earlier, deflation was first proposed in [38] but required OS and application cooperation, while we focus on a hypervisor-only deflation approach.…”

Section: Related Workmentioning

confidence: 99%

Cloud-scale VM-deflation for Running Interactive Applications On Transient Servers

Fuerst

Ali-Eldin

Shenoy

et al. 2020

Proceedings of the 29th International Symposium on High-Performance Parallel and Distributed Computing

View full text Add to dashboard Cite

Transient computing has become popular in public cloud environments for running delay-insensitive batch and data processing applications at low cost. Since transient cloud servers can be revoked at any time by the cloud provider, they are considered unsuitable for running interactive application such as web services. In this paper, we present VM deflation as an alternative mechanism to server preemption for reclaiming resources from transient cloud servers under resource pressure. Using real traces from top-tier cloud providers, we show the feasibility of using VM deflation as a resource reclamation mechanism for interactive applications in public clouds. We show how current hypervisor mechanisms can be used to implement VM deflation and present cluster deflation policies for resource management of transient and on-demand cloud VMs. Experimental evaluation of our deflation system on a Linux cluster shows that microservice-based applications can be deflated by up to 50% with negligible performance overhead. Our clusterlevel deflation policies allow overcommitment levels as high as 50%, with less than a 1% decrease in application throughput, and can enable cloud platforms to increase revenue by 30%.Before presenting our deflation techniques, we examine the efficacy and feasibility of deflating public cloud applications. We use publicly-available resource usage traces from two top-tier cloud providers, Azure [14] and Alibaba [15]. The goal of our analysis is to understand the feasibility of deflating CPU, memory, disk, and network allocations of real cloud applications, and specifically interactive web applications, under time-varying workloads that they exhibit. We seek to answer two key research questions through our feasibility analysis: (1) How much slack is present in cloud VMs and by how much can these VMs be safely deflated without any performance impact? (2) How does workload class and VM size impact the deflatability of VMs? Application Behavior under DeflationWe first present an abstract model to capture the performance behavior of an application under different amounts of resource Performance Slack Knee Linear Deflation %

show abstract

“…Smith et al [60] proposed a system for dynamically allocating memory amongst virtual machines at runtime, and they evaluated six allocation policies implemented within the system. Zhang et al [61] proposed iBalloon, a light-weight, accurate and transparent prediction based mechanism to enable more customizable and efficient ballooning policies for rebalancing memory resources among VMs.…”

Section: Related Workmentioning

confidence: 99%

Interdomain I/O Optimization in Virtualized Sensor Networks

Jiang

Fan

Qiu

et al. 2018

Sensors

View full text Add to dashboard Cite

In virtualized sensor networks, virtual machines (VMs) share the same hardware for sensing service consolidation and saving power. For those VMs that reside in the same hardware, frequent interdomain data transfers are invoked for data analytics, and sensor collaboration and actuation. Traditional ways of interdomain communications are based on virtual network interfaces of bilateral VMs for data sending and receiving. Since these network communications use TCP/IP (Transmission Control Protocol/Internet Protocol) stacks, they result in lengthy communication paths and frequent kernel interactions, which deteriorate the I/O (Input/Output) performance of involved VMs. In this paper, we propose an optimized interdomain communication approach based on shared memory to improve the interdomain communication performance of multiple VMs residing in the same sensor hardware. In our approach, the sending data are shared in memory pages maintained by the hypervisor, and the data are not transferred through the virtual network interface via a TCP/IP stack. To avoid security trapping, the shared data are mapped in the user space of each VM involved in the communication, therefore reducing tedious system calls and frequent kernel context switches. In implementation, the shared memory is created by a customized shared-device kernel module that has bidirectional event channels between both communicating VMs. For performance optimization, we use state flags in a circular buffer to reduce wait-and-notify operations and system calls during communications. Experimental results show that our proposed approach can provide five times higher throughput and 2.5 times less latency than traditional TCP/IP communication via a virtual network interface.

show abstract

iBalloon: Efficient VM Memory Balancing as a Service

Cited by 10 publications

References 11 publications

Memory-harvesting VMs in cloud platforms

Memory-harvesting VMs in cloud platforms

Cloud-scale VM-deflation for Running Interactive Applications On Transient Servers

Interdomain I/O Optimization in Virtualized Sensor Networks

Contact Info

Product

Resources

About