1 / 10

Virtual Batching for Performance Improvement and Energy Efficiency

Virtual Batching for Performance Improvement and Energy Efficiency. Dazhao , C heng 5/3/2012. Outline. Introduction and background. Why we do it? How we do it? Technique design of system. Experiment and evaluation. Power and performance in Data Center. Application performance.

salali
Download Presentation

Virtual Batching for Performance Improvement and Energy Efficiency

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Virtual Batching for Performance Improvement and Energy Efficiency Dazhao, Cheng 5/3/2012

  2. Outline Introduction and background. Why we do it? How we do it? Technique design of system. Experiment and evaluation.

  3. Power and performance in Data Center Application performance Power consumption Less hardware resource means less power consumption But better performance need more resource usage

  4. Why we do batching? • Most Web servers process requests in the order of arrival. (It may lead to low cache hit rate). • Batching requests in a content-aware manner can improve the cache hit rate of Web servers. • It provides the opportunity to improve the performance of multi-tier applications. Cache Clients Requests Response

  5. Batching mechanism (batcher) The main functions of this batcher are: 1.Requests collection; 2.Requests classification; 3.Requests reordering.

  6. System Architecture Power control loop Batching control loop

  7. System Implementation 1) We build a testbed consisting of one server and one client. The server is running CentOS 5.8 with Linux kernel 2.6.18. 2) We use RUBiSas the benchmark application for experiment. RUBiS provides a web auction application that modeled in a similar way of ebay.com. 3) We create three VMs by Xen 3.1, Apache web server in the first, PHP application server in the second, and MYSQL database server in the third. 4) Each VM is allocated 1 VCPU and 512 MB memory. All VMs use Ubuntu server 10.04 with Linux kernel 2.6.35.

  8. Evaluation The first scenario is a under-loaded (medium) system:

  9. Evaluation The secondscenario is a slightly overloaded system: Here we set the SLA of response time requirement is 1000 ms.

  10. Conclusion 1) In this work, we have proposed and developed a content-aware batching mechanism to improve the performance of applications. 2) Its main contributions are the precise control of batching interval length to improve performance and avoid SLA violations.

More Related