Faster Placement of Virtual Machines through Adaptive Caching

Einziger, Gil; Goldstein, Maayan; Sarar, Yaniv

doi:10.1109/infocom.2019.8737378

Cited by 15 publications

(13 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In (9), the first term is the fixed cost associated with activating VM m, which is incurred only if m is not already active (the summation can be at most 1, as per (3)). The second term is the proportional cost associated with the additional computation capability needed at VM m to guarantee stability, with ✏ being a positive, arbitrarily small value.…”

Section: A Steps 1-2: Bipartite Graph and Hungarian Algorithmmentioning

confidence: 99%

“…Importantly, significant cost savings can be achieved by sharing individual VNFs or sub-slices among services, whenever possible. The vast majority of VNF placement studies [6]- [9] consider scenarios where all placement decisions are made by a centralized entity, often the NFV Orchestrator (NFVO) in the ETSI Management and Orchestration (MANO) framework [5], [10]. Also, such an entity is in the position to make finegrained decisions on the usage of individual hosts and links.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Reducing Service Deployment Cost Through VNF Sharing

Malandrino

Chiasserini

Einziger

et al. 2019

IEEE/ACM Trans. Networking

Self Cite

View full text Add to dashboard Cite

Thanks to its computational and forwarding capabilities, the mobile network infrastructure can support several third-party ("vertical") services, each composed of a graph of virtual (network) functions (VNFs). Importantly, one or more VNFs are often common to multiple services, thus the services deployment cost could be reduced by letting the services share the same VNF instance instead of devoting a separate instance to each service. By doing that, however, it is critical that the target KPI (key performance indicators) of all services are met. To this end, we study the VNF sharing problem and make decisions on (i) when sharing VNFs among multiple services is possible, (ii) how to adapt the virtual machines running the shared VNFs to the combined load of the assigned services, and (iii) how to prioritize the services traffic within shared VNFs. All decisions aim to minimize the cost for the mobile operator, subject to requirements on end-to-end service performance, e.g., total delay. Notably, we show that the aforementioned priorities should be managed dynamically and vary across VNFs. We then propose the FlexShare algorithm to provide near-optimal VNF-sharing and priority assignment decisions in polynomial time. We prove that FlexShare is within a constant factor from the optimum and, using real-world VNF graphs, we show that it consistently outperforms baseline solutions.

show abstract

Section: A Steps 1-2: Bipartite Graph and Hungarian Algorithmmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Reducing Service Deployment Cost Through VNF Sharing

Malandrino

Chiasserini

Einziger

et al. 2019

IEEE/ACM Trans. Networking

Self Cite

View full text Add to dashboard Cite

show abstract

“…In Cloud context, the scheduler remains of interest without considerations of virtualization and its overheads [28] and [29]. The Openstack scheduler performance can be enhanced as per developer's view point discussed in [30] and [31]. Detailed comparisons of Openstack features with the other open source solutions such as OpenNebola and VMware are presented in [32]- [34], and [35] in detail.…”

Section: Related Workmentioning

confidence: 99%

Characterizing Dynamic Load Balancing in Cloud Environments Using Virtual Machine Deployment Models

et al. 2019

View full text Add to dashboard Cite

The ever growing computational demands of users call for efficient cloud resource management to avoid service-level agreement (SLA) violation. Virtualization co-locates multiple virtual machines (VMs) on a single physical server to share the underlying resources for efficient resource management. However, the decision about ''what'' and ''where'' to place workloads significantly impacts performance of hosted workloads. Existing cloud schedulers consider a single resource (RAM) to co-locate workloads that as a result lead to SLA violation due to non-optimal VM placement. To handle this issue, current study has updated nova scheduler to propose a multi-resource based VM placement approach to improve application performance in terms of central processing unit (CPU) utilization and execution time. Experimentally we have shown that our proposed method has lessened application execution time by 50% when compared with one of the well-known technique.

show abstract

“…A great deal of previous research into VM consolidation has focused on reducing the number of PMs in a cloud datacenter, while VM consolidation schemes often pursue different performance objectives [20,21]. Also, resource contention between colocated VMs and operational interference issues are not accurately investigated [7,22,23]. It is critical to evaluate the resilience to interference of migrating VM from other colocated VMs prior to consolidating them together on a same PM.…”

Section: Introductionmentioning

confidence: 99%

Design and evaluation of a joint profit and interference-aware VMs consolidation in IaaS cloud datacenter

RahimiZadeh

Dehghani

2021

Cluster Comput

View full text Add to dashboard Cite

Cloud service providers employ virtualization technologies in order to deliver energy efficiency and optimize resource utilization in cloud datacenters. In this way, Virtual Machine (VM) consolidation is an efficient technique for reducing energy related costs and environmental concerns. This technique consists of placing as much as possible VMs on as less as possible Physical Machines (PM)s by means of live VM migration. However, VM consolidation approaches bring advantages, they cannot efficiently guarantee performance isolation of co-located VMs resulting in operational interference issues. Therefore, it is necessary to be aware of workload performance degradation issues during VMs consolidation in a datacenter. In this paper, we propose a novel joint profit and interference-aware scheduling scheme (PIAS) to efficiently consolidate VMs that host multi-tier application workloads in Infrastructure-as-a-Service (IaaS) cloud datacenter. The PIAS scheme considers profit and cost of power consumption, operational interference of VMs, resource utilization, and service level agreements (SLA)s as well as amount of transferred memory pages during live VM migration for consolidating VMs. In this regard, a stochastic dynamic programming is introduced to model the operational behavior of VM. Also, an optimization problem is introduced to maximize provider's profit and minimize the overall execution cost of application workloads. We simulate a series of experiments based on real workloads traces for different IaaS cloud datacenter configurations. The PIAS scheme on average provides better results than other competitors and the results demonstrate how our approach at least improves profit, energy efficiency, and service down time by 40%, 29%, and 35%, respectively.

show abstract

Faster Placement of Virtual Machines through Adaptive Caching

Cited by 15 publications

References 20 publications

Reducing Service Deployment Cost Through VNF Sharing

Reducing Service Deployment Cost Through VNF Sharing

Characterizing Dynamic Load Balancing in Cloud Environments Using Virtual Machine Deployment Models

Design and evaluation of a joint profit and interference-aware VMs consolidation in IaaS cloud datacenter

Contact Info

Product

Resources

About