Home Strategy Activities Grants Publications People Sponsors Blog Contact Us 
  

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

publications:publi:rhjwshcsszjzjv2015icrtlcis [2019/10/25 07:54] (current)
Line 1: Line 1:
  
 +<​html><​div id="​bib">​
 +
 +<p> <h1> RhJwShCsSzJzJv2015icrtlcis</​h1>​
 + <​p><​span class="​BibAuthor">​R. Han, J. Wang, S. Huang, C. Shao, S. Zhan, J. Zhan, J.L. Vazquez-Poletti</​span>​. <span class="​BibInProceedingsTitle">​Interference-Aware Component Scheduling for Reducing Tail Latency in Cloud Interactive Services</​span>​. In <span class="​BibInProceedingsBooktitle">​35th IEEE International Conference on Distributed Computing Systems (ICDCS 2015)</​span>,​ Pages 744-745, 2015.</​P><​p>​
 +<p>
 +<a name="​abstract"></​a><​h2>​ Abstract ​ </h2> <​P> ​
 +Large-scale interactive services usually divide requests into multiple sub-requests and distribute them to a large number of server components for parallel execution. Hence the tail latency (i.e. The slowest component'​s latency) of these components determines the overall service latency. On a cloud platform, each component shares and competes node resources such as caches and I/O bandwidths with its co-located jobs, hence inevitably suffering from their performance interference. In this paper, we study the short-running jobs in a 12k-node Google cluster to illustrate the dynamic resource demands of these jobs, resulting in both individual components'​ latency variability over time and across different nodes and hence posing a major challenge to maintain low tail latency. Given this motivation, this paper introduces a dynamic and interference-aware scheduler for large-scale,​ parallel cloud services. At each scheduling interval, it collects workload and resource contention information of a running service, and predicts both the component latency on different nodes and the overall service performance. Based on the predicted performance,​ the scheduler identifies straggling components and conducts near-optimal component-node allocations to adapt to the changing workloads and performance interferences. We demonstrate that, using realistic workloads, the proposed approach achieves significant reductions in tail latency compared to the basic approach without scheduling<​p>​
 + <a name="​keyword"></​a>​ <​h2>​Keywords </h2> <p> [ <a href="/​doku.php?​id=publications:​keyword:​tin2012-31518">​Tin2012-31518 </a> ] [ <a href="/​doku.php?​id=publications:​keyword:​cloud">​ Cloud</​a>​ ] 
 +<a name="​contact"></​a><​h2>​ Contact ​ </h2> <​P> ​
 +<a href="​mailto:​jlvazquez@fdi.ucm.es">​Jose Luis  Vazquez-Poletti</​a> ​ <a href="/​jlvazquez">​ <img src="/​lib/​exe/​fetch.php?​w=&​h=&​cache=cache&​media=html_icon.png"​ align=top border=0 alt =""></​a><​br> ​
 +
 +<a name="​bib"></​a><​h2>​ BibTex Reference ​ </h2> <​P> ​
 +@InProceedings{RhJwShCsSzJzJv2015icrtlcis,​ <​br>&​nbsp;&​nbsp;&​nbsp;​Author = {Han, R. and Wang, J. and Huang, S. and Shao, C. and Zhan, S. and Zhan, J. and Vazquez-Poletti,​ J.L.},<​br>&​nbsp;&​nbsp;&​nbsp;​Title = {Interference-Aware Component Scheduling for Reducing Tail Latency in Cloud Interactive Services},<​br>&​nbsp;&​nbsp;&​nbsp;​BookTitle = {35th IEEE International Conference on Distributed Computing Systems (ICDCS 2015)},<​br>&​nbsp;&​nbsp;&​nbsp;​Pages = {744--745},<​br>&​nbsp;&​nbsp;&​nbsp;​Publisher = {IEEE},<​br>&​nbsp;&​nbsp;&​nbsp;​Year = {2015}<​br>​} <​br><​p>​
 +
 +</​div></​html>​
Admin · Log In