Adaptive Prediction Models for Data Center Resources Utilization Estimation

Baig, Shuja-ur-Rehman; Iqbal, Waheed; Berral, Josep Lluís; Erradi, Abdelkarim; Carrera, David

doi:10.1109/tnsm.2019.2932840

Cited by 62 publications

(18 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This is a common characteristic of data source nodes as they usually host services whose workloads change over time. As a consequence, several large-scale web services (e.g., Alibaba and Wikipedia) and ML inference services experience bursty request loads in the order of minutes [30]- [33]. These services require variable amount of compute resource to meet their SLAs despite changing request loads.…”

Section: B Adaptive Query Partitioning In Query Enginesmentioning

confidence: 99%

Jarvis: Large-scale Server Monitoring with Adaptive Near-data Processing

Sandur¹,

Park²,

Volos³

et al. 2022

Preprint

View full text Add to dashboard Cite

Rapid detection and mitigation of issues that impact performance and reliability is paramount for large-scale online services. For real-time detection of such issues, datacenter operators use a stream processor and analyze streams of monitoring data collected from servers (referred to as data source nodes) and their hosted services. The timely processing of incoming streams requires the network to transfer massive amounts of data, and significant compute resources to process it. These factors often create bottlenecks for stream analytics.To help overcome these bottlenecks, current monitoring systems employ near-data processing by either computing an optimal query partition based on a cost model or using modelagnostic heuristics. Optimal partitioning is computationally expensive, while model-agnostic heuristics are iterative and search over a large solution space. We combine these approaches by using model-agnostic heuristics to improve the partitioning solution from a model-based heuristic. Moreover, current systems use operator-level partitioning: if a data source does not have sufficient resources to execute an operator on all records, the operator is executed only on the stream processor. Instead, we perform data-level partitioning-i.e., we allow an operator to be executed both on a stream processor and data sources.We implement our algorithm in a system called Jarvis, which enables quick adaptation to dynamic resource conditions. Our evaluation on a diverse set of monitoring workloads suggests that Jarvis converges to a stable query partition within seconds of a change in node resource conditions. Compared to current partitioning strategies, Jarvis handles up to 75% more data sources while improving throughput in resource-constrained scenarios by 1.2-4.4×.

show abstract

Section: B Adaptive Query Partitioning In Query Enginesmentioning

confidence: 99%

Jarvis: Large-scale Server Monitoring with Adaptive Near-data Processing

Sandur¹,

Park²,

Volos³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Many studies have been undertaken that have proposed estimation methods based on traditional machine learning approaches to generate predictions in adaptive and dynamic ways for the estimation of resource utilization in data centers in a cloud computing context [41][42][43][44].…”

Section: Related Workmentioning

confidence: 99%

A deep learning-based resource usage prediction model for resource provisioning in an autonomic cloud computing environment

Al-Asaly

Bencherif

Alsanad

et al. 2021

Neural Comput & Applic

View full text Add to dashboard Cite

Cloud computing enables clients to acquire cloud resources dynamically and on demand for their cloud applications and services. For cloud providers, especially, Software as a Service (SaaS) providers, the prediction of future cloud resource requirements, such as CPU usage for their cloud applications, to implement client requests is a complex task because it depends on incoming workloads. Due to workload fluctuations, it is difficult for SaaS cloud providers to predict or forecast future demand for resource usage in the next time interval and, accordingly, to allocate the required resources. Furthermore, cloud computing systems consist of many virtual machines (VMs), which increases the complexity of the prediction problem due to the correlations that exist between the large workload data in these VMs. Therefore, accurate resource usage forecasting remains a challenge, and relatively few studies have explored the prediction of CPU usage for VMs in cloud data centers. This paper proposes an autonomic and intelligent workload forecasting method for cloud resource provisioning based on the concept of autonomic computing and a deep learning approach. In particular, to predict future demand for CPU usage and determine how to respond to workload fluctuations in the next interval, we propose an efficient deep learning model based on a diffusion convolutional recurrent neural network (DCRNN). Existing deep learning models that are widely applied cannot handle accurate real-time forecasting due to the presence of inconsistent and nonlinear workloads in cloud computing systems. The goal of the proposed deep learning model is to improve forecasting accuracy and minimize the error between the predicted and the actual workloads. The effectiveness of the proposed DCRNN-based deep learning model was evaluated using experiments on a real-world dataset of PlanetLab's CPU usage traces. The results indicate that the proposed approach outperformed other existing deep learning models, achieving a mean absolute percentage error of 0.18 and root-mean-square error of 2.40.

show abstract

“…On the other hand, an inaccurate estimation leads to either over or under-provisioning of data center resources, as depicted in Figure 4, resulting into wastage of resources, unnecessary power consumption and violations of the service level agreement. A leading challenge in workload forecasting is the presence of multi-tenant co-hosted applications characterized by nonlinear, dynamic and time-varying nature [10]. For instance, at any time, millions of requests could be generated, whereas at the next instance of time, very few requests or even none might be issued, resulting in sudden peaks and rock bottoms in workload patterns.…”

Section: A Workload Forecastingmentioning

confidence: 99%

“…Therefore, a workload forecasting method becomes a significant research problem where the estimation strategy must be capable of accurately estimating future resource needs while adapting to dynamic workload demands in a data center environment. Literature: Iqbal et al proposed a method that can adap-tively and automatically identify the appropriate model for resource utilization estimation [10]. It trains a classifier through different scenarios and a corresponding resource estimator for each, in order to learn the best regression model to produce the workload prediction.…”

Section: A Workload Forecastingmentioning

confidence: 99%

Next-Generation Data Center Network Enabled by Machine Learning: Review, Challenges, and Opportunities

Dong¹,

Munir

Tout

et al. 2021

IEEE Access

View full text Add to dashboard Cite

Data center network (DCN) is the backbone of many emerging applications from smart connected homes to smart traffic control and is continuously evolving to meet the diverse and ever-increasing computing requirements of these applications. The data centers often have tens of thousands of components such as servers and switches/routers that work together to achieve a common objective and serve these applications. Managing such large data centers is a tedious process and demands for automation, intelligent control and decision making within the data center. Recently both the industry and academia have focused on bringing intelligence to the control, automation and management of DCNs. Despite the variety of works that surveyed ML for networking, to the best of our knowledge, none has focused on DCN, which makes this survey original. Readers in the academic and industrial communities will all benefit from a comprehensive discussion of the ML solutions applied in DCN to address critical essential problems, including workload forecasting, traffic control and optimization, topology management, network state prediction and failures analysis, and security. Furthermore, this article outlines the challenges and concludes with the future research venues in adopting ML for automatic, intelligent and autonomous DCNs.

show abstract

Adaptive Prediction Models for Data Center Resources Utilization Estimation

Cited by 62 publications

References 36 publications

Jarvis: Large-scale Server Monitoring with Adaptive Near-data Processing

Jarvis: Large-scale Server Monitoring with Adaptive Near-data Processing

A deep learning-based resource usage prediction model for resource provisioning in an autonomic cloud computing environment

Next-Generation Data Center Network Enabled by Machine Learning: Review, Challenges, and Opportunities

Contact Info

Product

Resources

About