The new tool enables a comprehensive assessment of data center performance

تتيح الأداة الجديدة إجراء تقييم شامل لأداء مركز البيانات

SDCBench Overview. attributed to him: intelligent computing (2022). DOI: 10.34133/2022/9810691

Data centers are used by countless other businesses, organizations, and operations, including large-scale online services such as e-commerce, search engines, online maps, social media, advertising, and more. These data centers map workloads, which includes sharing data center resources to improve server utilization.

However, this can degrade performance. To study this problem and find potential solutions, researchers need tools to assess the co-location of the workload. While these tools have been developed before, they may only measure one aspect at the expense of other factors, which limits their usefulness.

A team of researchers from Tianjin University and Dalian University of Technology, both in China, have developed a cross-site benchmarking suite of workloads, called SDCBench, to address previous problems and provide a comprehensive analysis.

Research published in intelligent computing On September 7th.

“The shared location of the workload can cause performance overlap that can degrade the performance of cloud services unpredictably, which not only reduces user experience It also hurts the resource efficiency of data centers, said corresponding author Liping Zhao, assistant professor in the Tianjin Key Laboratory for Advanced Networks in the College of Intelligence and Computing at Tianjin University, China.

To get around this problem, researchers are trying to boost the ability of isolation – which indicates privacy concerns Related to resource sharing in data centers – to cloud systems through hardware and software approaches. However, proposed solutions may require new software or hardware updates, which some cloud providers cannot or will not provide.

“The need for predictable service performance in data centers brings new challenges and opportunities for cloud system design that seeks to optimize resource utilization at the server level but does not compromise performance at the application level,” Zhao said.

“Unfortunately, the lack of a comprehensive set of co-location standards for workloads makes studying this emerging issue difficult. The co-location standard can help cloud service providers understand and improve the potential for isolating infrastructures, thereby increasing their adoption by cloud users. “

The researchers developed SDCBench, a cross-site benchmarking suite for workloads that includes 16 latency critical services and applications that cover a wide range of cloud scenarios—which means there should be very little delay in response time.

“SDCBench enables cloud tenants to understand the performance isolation capacity of data centers and to choose the best one cloud servicesZhao said. “For cloud service providers, this also helps them improve quality of service to increase their revenue.”

In conjunction with the introduction of the new set of criteria, the researchers proposed the concept of latency entropy, which is inspired by the physical definition of entropy to mean the degree of disorder within a system, to measure uncertainty in cloud systems.

“When contention over shared resources occurs between different applications, system behavior becomes unorganized and unpredictable,” Zhao said. “To help users understand application performance changes using observable metrics, SDCBench defines latency entropy that describes variations in tail latency to measure a system’s isolation ability.”

Researchers have shown that SDCBench can simulate various cloud scenarios by locating workloads with simple configurations. They also evaluated and compared the latency entropy of major cloud service providers using their own benchmarking tool.

According to Zhao, one of the most exciting aspects of the research is that the SDC Bench and the comprehensive framework built on it are publicly available.

“We have implemented a comprehensive evaluation framework based on SDCBench that can automatically configure, deploy, and evaluate applications on cloud platforms. This framework is open source and can be easily extended to new cloud systemsZhao said.

more information:
Yan’an Yang et al., SDCBench: A reference set for compiling and evaluating workload in data centers, Available Here. intelligent computing (2022). DOI: 10.34133/2022/9810691

github: github.com/TankLabTJU/sdcbench/tree/sdcbench-v2.0/

Introduction of intelligent computing

the quote: New Tool Enables Comprehensive Assessment of Data Center Performance (2022, November 17), Retrieved November 17, 2022 from https://techxplore.com/news/2022-11-tool-enables-compuate-datacenter.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.

Leave a Comment