iBench: Quantifying interference for datacenter applications

Delimitrou, Christina; Kozyrakis, Christos

doi:10.1109/iiswc.2013.6704667

Cited by 110 publications

(43 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…AutoPro uses a PI controller to asymptotically close this gap and can work with any metric (eg, frame/s) as long as developers can provide it. Delimitrou and Kozyrakis [13] proposed iBench to inject/run a series of 15 snippets into a VM to pressure its running application and identify the amount of contention it can tolerate before violating its QoS.…”

Section: Agent Based Approachesmentioning

confidence: 99%

“…vmDataCollector(vmName,startTime,finishTime) collects/records performance of a VM under imposed limitations through polling several Hypervisor level metrics (eg, CPU utilization): it neither demands nor needs any specific metric from the VM itself. This makes vmBBProfiler unique when compared with many other similar approaches that either ask developers to provide specific metrics to reflect a VM's performance [5,6,8,[11][12][13]21], or use peripheral systems to collect external metrics to reflect such performance [24]. Because, the vmBBProfiler remains fully agnostic to the actual internal processes of a VM, we refer to our approach as a 'Black-Box' technique.…”

Section: Architecture Of Vmbbprofilermentioning

confidence: 99%

See 1 more Smart Citation

vmBBProfiler: a black-box profiling approach to quantify sensitivity of virtual machines to shared cloud resources

2017

View full text Add to dashboard Cite

Virtualized Data Centers are packed with numerous web and cloud services nowadays. In such large infrastructures, providing reliable service platforms depends heavily on efficient sharing of physical machines (PMs) by virtual machines (VMs). To achieve efficient consolidation, performance degradation of co-located VMs must be correctly understood, modeled, and predicted. This work is a major step toward understanding such baffling phenomena by not only identifying, but also quantifying sensitivity of general purpose VMs to their demanded resources. vmBBProfiler, our proposed system in this work, is able to systematically profile behavior of any general purpose VM and calculate its sensitivity to system provided resources such as CPU, Memory, and Disk. vmBBProfiler is evaluated using 12 well-known benchmarks, varying from pure CPU/Mem/Disk VMs to mixtures of them, on three different PMs in our VMware-vSphere based private cloud. Extensive empirical results conducted over 1200 h of profiling prove the efficiency of our proposed models and solutions; it also opens doors for further research in this area.

show abstract

Section: Agent Based Approachesmentioning

confidence: 99%

Section: Architecture Of Vmbbprofilermentioning

confidence: 99%

vmBBProfiler: a black-box profiling approach to quantify sensitivity of virtual machines to shared cloud resources

2017

View full text Add to dashboard Cite

show abstract

“…Interfering Workload: Interfering workload anomalies are the major cause of performance degradation in data centers (DC) [17]. Interfering workload anomalies results from competition for resources and occur due to various reasons; as simple as unannounced maintenance on a cluster (e.g., security scans), or a storage array that is performing a system operation such as replication and RAID construction.…”

Section: ) Anomaliesmentioning

confidence: 99%

Detecting Discontinuities in Large Scale Systems

Malik

Davis

Godfrey

et al. 2014

2014 IEEE/ACM 7th International Conference on Utility and Cloud Computing

View full text Add to dashboard Cite

Abstract-Cloud providers and data centers rely heavily on forecasts to accurately predict future workload. This information helps them in appropriate virtualization and cost-effective provisioning of the infrastructure. The accuracy of a forecast greatly depends upon the merit of performance data fed to the underlying algorithms. One of the fundamental problems faced by analysts in preparing data for use in forecasting is the timely identification of data discontinuities. A discontinuity is an abrupt change in a time-series pattern of a performance counter that persists but does not recur. Analysts need to identify discontinuities in performance data so that they can a) remove the discontinuities from the data before building a forecast model and b) retrain an existing forecast model on the performance data from the point in time where a discontinuity occurred. There exist several approaches and tools to help analysts identify anomalies in performance data. However, there exists no automated approach to assist data center operators in detecting discontinuities in the first place. In this paper, we present and evaluate our proposed approach to help data center analysts and cloud providers automatically detect discontinuities. A case study on the performance data obtained from a large cloud provider and performance tests conducted using an open source benchmark system show that our proposed approach provides on average precision of 84% and recall 88%. The approach doesn't require any domain knowledge to operate.

show abstract

“…RU quality: The utility an application can extract from an RU depends on the hardware type (e.g., 2 GHz vs 3 GHz core) and the interference on shared resources from other jobs on the same server. The scheduler [12,13] obtains the interference preferences of an incoming job using a small set of microbenchmarks to inject pressure of increasing intensity (from 0 to 99%) on one of ten shared resources of interest [11]. Interference preferences capture, first, the amount of pressure t i a job can tolerate in each shared resource i ∈ [1, N ], and second, the amount of pressure c i it itself will generate in that resource.…”

Section: Analytical Frameworkmentioning

confidence: 99%

Tarcil

Delimitrou

Sánchez

Kozyrakis

2015

Proceedings of the Sixth ACM Symposium on Cloud Computing

Self Cite

113

View full text Add to dashboard Cite

Scheduling diverse applications in large, shared clusters is particularly challenging. Recent research on cluster scheduling focuses either on scheduling speed, using sampling to quickly assign resources to tasks, or on scheduling quality, using centralized algorithms that search for the resources that improve both task performance and cluster utilization.We present Tarcil, a distributed scheduler that targets both scheduling speed and quality. Tarcil uses an analytically derived sampling framework that adjusts the sample size based on load, and provides statistical guarantees on the quality of allocated resources. It also implements admission control when sampling is unlikely to find suitable resources. This makes it appropriate for large, shared clusters hosting shortand long-running jobs. We evaluate Tarcil on clusters with hundreds of servers on EC2. For highly-loaded clusters running short jobs, Tarcil improves task execution time by 41% over a distributed, sampling-based scheduler. For more general scenarios, Tarcil achieves near-optimal performance for 4× and 2× more jobs than sampling-based and centralized schedulers respectively.

show abstract

iBench: Quantifying interference for datacenter applications

Cited by 110 publications

References 32 publications

vmBBProfiler: a black-box profiling approach to quantify sensitivity of virtual machines to shared cloud resources

vmBBProfiler: a black-box profiling approach to quantify sensitivity of virtual machines to shared cloud resources

Detecting Discontinuities in Large Scale Systems

Tarcil

Contact Info

Product

Resources

About