Recently, size-based policies such as SRPT and FSP have been proposed for scheduling requests in web servers. SRPT and FSP are superior to policies that ignore request size, such as PS, in both efficiency and fairness given heavy-tailed service times. However, a central assumption that is usually made in implementing size-based policies in a web server is that the service time of a request is strongly correlated with the size of the file it serves. This paper shows how the performance of SRPT and FSP are affected by the degree of this correlation. We developed a simulator that supports both M/G/1/m and G/G/n/m queuing models. The simulator can be driven with trace data, which can be taken from the logs of modified Apache servers, or which can be produced by a workload generator we have developed that allows us to control the correlation. Using both trace data and generated data, we find that the degree of correlation has a dramatic effect on the performance of SRPT and FSP. In response, we propose and evaluate domain-based scheduling, a simple technique that better estimates connection times by making use of the source IP address of the request. Domain-based scheduling improves SRPT and FSP performance on web servers, particularly in regimes where correlation is low, thus making size-based policies such as these more broadly deployable.
|
1087
|
Traffic: The Failure of Poisson Modeling
– Paxson, Floyd
- 1995
|
|
872
|
Self-similarity in world wide web traffic: evidence and possible causes
– Crovella, Bestavros
- 1997
|
|
557
|
Generating Representative Web Workloads for Network and Server Performance Evaluation
– Barford, Crovella
- 1998
|
|
452
|
End-to-end routing behavior in the Internet
– Paxson
- 1997
|
|
200
|
Dynamically forecasting network performance using the network weather service
– Wolski
- 1998
|
|
185
|
SPAND: Shared passive network performance discovery
– SESHAN, STEMM, et al.
- 1997
|
|
146
|
On the Constancy of Internet Path Properties
– Zhang, Duffield, et al.
- 2001
|
|
103
|
Analysis of SRPT scheduling: investigating unfairness
– Bansal, Harchol-Balter
- 2001
|
|
91
|
Analyzing Stability in Wide-Area Network Performance
– BALAKRISHNAN, SESHAN, et al.
- 1997
|
|
90
|
Web Facts and Fantasy
– Manley, Seltzer
- 1997
|
|
90
|
The Available Capacity of a Privately Owned Workstation Environment
– Mutka, Livny
- 1991
|
|
90
|
Queueing theory with Computer Science Applications
– Probability
- 1990
|
|
70
|
ICP and the Squid web cache
– Wessels, Claffy
|
|
66
|
The case for cooperative networking
– Padmanabhan, Sripanidkulchai
- 2002
|
|
56
|
A proof of the optimality of the shortest remaining processing time discipline
– Schrage
- 1968
|
|
54
|
Size-based scheduling to improve web performance
– Harchol-Balter, Schroeder, et al.
- 2003
|
|
48
|
Collecting network status information for networkaware applications
– Miller, Steenkiste
- 2000
|
|
46
|
Performance characteristics of mirror servers on the internet
– MYERS, DINDA, et al.
- 1999
|
|
45
|
The Queue M/G/1 with the Shortest Remaining Processing Time Discipline
– Schrage, Miller
- 1966
|
|
40
|
et.al., “Implementation of SRPT scheduling in web servers
– Harchol-Balter
- 2001
|
|
38
|
Dynamics of internet routing information
– CHINOY
- 1993
|
|
37
|
O’Halaron. An evaluation of linear models for host load prediction
– Dinda, R
- 1999
|
|
32
|
Web servers under overload: How scheduling can help
– Schroeder, Harchol-Balter
- 2003
|
|
27
|
Operating System Concepts (5th Edition
– SILBERSCHATZ, GALVIN
- 1998
|
|
26
|
O.: Probability, Statistics, and Queueing Theory with Computer Science Applications, 2d ed
– Allen
- 1990
|
|
24
|
Measuring Web performance in the wide area
– Barford, Crovella
- 1999
|
|
24
|
Wide-AreaTraffic: The Failure of Poisson Modeling
– Paxson, S
- 1994
|
|
16
|
The variance of delay time in queueing system M/G/1 with optimal strategy SRPT,” AEU, Archiv fuer Elektronik und Uebertragungstechnik
– Perera
- 1993
|
|
14
|
Numerical Methods for Fitting and Simulating Autoregressive-to-Anything Processes
– CARIO, NELSON
- 1998
|
|
14
|
Fairness and efficiency in web server protocols
– Friedman, Henderson
- 2003
|
|
12
|
Empirical model of WWW document arivals at access links
– Deng
- 1996
|
|
12
|
RFC 1519: Classless inter-domain routing (CIDR): an address assignment and aggregation strategy
– Fuller, Li, et al.
- 1993
|
|
12
|
Properties and applications of the optimal queueing strategy SRPT - a survey. Archiv fur Elektronik und Uebertragungstechnik
– SCHREIBER
- 1993
|
|
10
|
The case for SRPT scheduling in web servers
– Harchol-Balter, Crovella, et al.
- 1998
|
|
9
|
Quantifying the properties of SRPT scheduling
– Gong, Williamson
- 2003
|
|
8
|
Size-based scheduling policies with inaccurate scheduling information
– LU, SHENG, et al.
- 2004
|
|
8
|
Web facts and fantasy
– Manly, Seltzer
- 1997
|
|
7
|
RFC 2186: Internet cache protocol (ICP), version 2. The Internet Engineering Taskforce
– WESSELS, CLAFFY
- 1997
|
|
7
|
Simulation evaluation of hybrid SRPT scheduling policies
– Gong, Williamson
- 2004
|
|
5
|
Analysis of a local-area bus system with controlled access
– BUX
- 1983
|
|
4
|
Queueing systems with long-range dependent input process and subexponential service times
– Xia, Liu
- 2003
|
|
3
|
rfc1518) An architecture for IP address allocation with CIDR
– REKHTER, LI
- 1993
|
|
2
|
a unix file system benchmark. http://www.textuality.com/bonnie
– Bonnie
|
|
2
|
Heavy-traffic analysis for the G/G/1 queue with heavy-tailed distributions. Queueing Systems 33
– BOXMA, COHEN
- 1999
|
|
1
|
rfc1517) Applicability statement for the implementation of Classes Inter-Domain Routing (CIDR
– HINDEN
- 1993
|
|
1
|
On the approximation of the mean response times of priority classes in a stable G/G/C/PR queue
– N, D
- 1992
|
