1 scheduling in server farms mor harchol-balter computer science dept carnegie mellon university...

1

Scheduling in Server FarmsScheduling in Server Farms

Mor Harchol-BalterComputer Science Dept

Carnegie Mellon University

[email protected]

2

Outline

I. Review of scheduling in single-server

FCFS

FCFSRouter

II. Supercomputing

III. Web server farm model

PSRouter

PS

IV. Towards Optimality …

SRPT

SRPT

SRPTRouter&

= today

= tomorrow

Metric: Mean

Response Time, E[T]

3

Single Server Model (M/G/1)

Poissonarrival processw/rate

Load E[X]<1

• CPU Lifetimes of UNIX jobs [Harchol-Balter, Downey 96]• Supercomputing job sizes [Schroeder, Harchol-Balter 00]• Web file sizes [Crovella, Bestavros 98, Barford, Crovella 98]• IP Flow durations [Shaikh, Rexford, Shin 99]

Job sizes with huge variance are everywhere in CS:

Top-heavy:top 1% jobsmake up halfload

Huge Variability

1

½

¼

xx 1}size JobPr{ ~

k x p

Bounded Pareto

X: job size (service requirement)

D.F.R.2

2

( )

[ ]X

Var XC

E X

28 50 is typicalXC

4

Scheduling Single Server (M/G/1)

Question: Order these scheduling policies for mean response time, E[T]:

Poissonarrival process

Load <1

1. FCFS (First-Come-First-Served, non-preemptive)

2. PS (Processor-Sharing, preemptive)

3. SJF (Shortest-Job-First, a.k.a., SPT, non-preemptive)

4. SRPT (Shortest-Remaining-Processing-Time, preemptive)

5. LAS (Least Attained Service, a.k.a., FB, preemptive)

Huge Varianc

e

5

Scheduling Single Server (M/G/1)Poissonarrival process

Load <1 Huge Varianc

e

LOW E[T] HIGH E[T]

SRPT < LAS < PS < SJF < FCFS

Surprisingly bad:(E[X2] term)

Requires D.F.R. [Righter,

Shanthikumar89]

OPT for allarrival

sequences[Schrage 67]

No “Starvation!” Even the biggest jobs prefer SRPT to PS:[Bansal, Harchol-Balter 01], [Wierman, Harchol-Balter 03]:

THM: E[T(x)]SRPT < E[T(x)]PS for all x, for Bounded Pareto, < .9.

~E[X2](shorts caughtbehindlongs)

Insensitiveto E[X2]

6

SRPT

FCFS SJF

24

LASLAS

FCFS SJF

20

LAS

FCFS SJF

16

LAS

FCFS SJF

12

FCFS SJF

LAS

8

Effect of Variability

Bounded Pareto job sizes

PS

C2 =

E[T]

7

Closed vs. Open Systems

QUESTION:What’s the

effect of scheduling?

Open System

Closed System

Send Receive

Think

8

Open System Results

E[T]FCFS

SRPT

Closed vs. Open Systems

Closed System Results

E[T]

FCFS

SRPT

Closed & open systems analyzed under same job size distribution,with same average load.

[Schroeder, Wierman, Harchol-Balter, NSDI 06]

9

Summary Single-Server

LESSONS LEARNED: Smart scheduling greatly improves mean response time. Variability of job size distribution is key. Closed system sees much reduced effect.

X: job size -- highly-variable

<1

Single-server system

10

Multiserver Model

Sched. policy

Router

Server farms: + Cheap + Scalable capacity

Sched. policy

Sched. policyRouting

(assignment)policyIncoming

jobs:Poisson Process

2 Policy Decisions

(Sometimes scheduling policy is fixed – legacy system)

11

Outline


FCFS

FCFSRouter

II. Supercomputing


PSRouter

PS


SRPT

SRPT

SRPTRouter&

Metric: Mean

Response Time, E[T]

12

Supercomputing Model

FCFS

Router

FCFS

FCFSRouting

(assignment)policy

Poisson Process

Jobs are not preemptible.

Assume hosts are identical. Jobs i.i.d. ~ G: highly variable size distribution.

Jobs processed in FCFS order.

Size may or may not be known. Initially assume known.

13

Q: Compare Routing Policies for E[T]?

FCFS

Router

FCFS

FCFS

Routingpolicy

Poisson Process

Jobs i.i.d. ~ G: highly variable

Supercomputing1. Round-Robin

2. Join-Shortest-Queue Go to host w/ fewest # jobs.

3. Least-Work-Left, equivalent to M/G/k/FCFS Go to host with least total work.

4. Central-Queue-Shortest-Job (M/G/k/SJF)

Host grabs shortest job when free.

5. Size-Interval Splitting Jobs are split up by size among hosts.

14

1. Round-Robin






FCFS

Router

FCFS

FCFS

Routingpolicy

Highly variable job sizes

Supercomputing

A: Size-Interval Splitting: best so farHighE[T]

LowE[T]

[Harchol-Balter, Crovella, Murta, JPDC 99]

15

1. Round-Robin






HighE[T]

LowE[T]

Central-Queue:+ Good utilization of servers.+ Some isolation for smalls

[Harchol-Balter, Crovella,Murta, JPDC 99].

Size-Interval WAY Better!- Worse utilization of servers.+ Great isolation for smalls!

Routing Policies: Remarks

16

Size-Interval Splitting

S

L

XL

M

Size-IntervalRouting

job size x

x f x ( )

Question: How to choose the size cutoffs? “To Balance Load or Not to Balance Load?”

S M L XL

dxxxfdxxxfdxxxfdxxxf )()()()( ? ? ?

17

Size-Interval Splitting

job size x

x f x ( )

Answer: Recent Research for case of Bounded Pareto job size: Pr{X>x} ~ x-

<1UNBALANCEfavor smalls

UNBALANCEfavor larges

BALANCE LOAD

=1 >1

[Harchol-Balter,Vesilo, 06+], [Glynn, Harchol-Balter, Ramanan, 06+]

L

S


FCFS

FCFS

ssss

LLL

18

Beyond Size-Interval Splitting

job size x

x f x ( )

Q: Is Size-Interval Splitting as good as it gets?

L

S


FCFS

FCFS

ssss

LLL

19

Size-Interval Splitting with Stealing

Answer: Allow Cycle Stealing!

L

S

Size-IntervalRoutingwith CycleStealing

FCFS

FCFS

Send Shorts Here

Send Longs Here.But, if idle, send Short.

Gain to Shorts is high Pain to Longs is very small.

Cycle Stealing analysis very hard: Fayolle, Iasnogorodski, Konheim,Meilijson, Melkman, Cohen, Boxma, van Uitert, Jelenkovic, Foley, McDonald,Harrison, Borst, Williams …

New easy approach: Dimensionality Reduction 2D1D [Harchol-Balter, Osogami, Scheller-Wolf, Squillante SPAA03]

20

1. Round-Robin






What if Don’t Know Job Size?

Q: What can we do to minimize E[T] when don’t know job size?

FCFS

Router

FCFS

FCFS

Routingpolicy

Highly variable job sizes

21

The TAGS algorithm“Task Assignment by Guessing Size”

Answer:When job reaches size limit for host, then itis killed and restarted from scratch at next host.

m

s

OutsideArrivals

Host 1

Host 2

Host 3

[Harchol-Balter, JACM 02]

22

Lowervariability

Highvariability

Results of Analysis

TAGS

Random

Least-Work-Left

23

Summary – Part I

LESSONS LEARNED: Smart scheduling greatly improves mean response time. Variability of job size distribution is key. Closed system sees much reduced effect.

LESSONS LEARNED: Greedy routing policies, like JSQ, LWL are poor. To combat variability, need size-interval splitting. By isolating smalls, can achieve effects of smart single-server policies Load UN-balancing Don’t need to know size.

X: job size -- highly-variable

<1

Single-server system

FCFS

FCFSRouter

Supercomputing

24

Tomorrow …

I. Review of scheduling in single-serverM/GI/1

FCFS

FCFSRouter

II. Supercomputing/Manufacturing


PS

PSRouter


SRPT

SRPT

SRPTRouter&

Homework

25

Scheduling in Multiserver SystemsPART II

Scheduling in Multiserver SystemsPART II

Mor Harchol-BalterComputer Science Dept

Carnegie Mellon University

[email protected]

26

Outline

FCFS

FCFSRouter


II. Supercomputing


PSRouter

PS


SRPT

SRPT

SRPTRouter&

27

Outline

FCFS

FCFSRouter


II. Supercomputing


PSRouter

PS


SRPT

SRPT

SRPTRouter&

28

Web Server Farm Model

HTTP requests are immediately dispatched to server. Requests are fully preemptible. Commodity servers utilized Do Processor-Sharing. Jobs i.i.d. ~ G: highly variable size distribution, 7 orders magnitude difference in job size [Crovella, Bestavros 98].

Router

Routingpolicy

Poisson Process

PS

PS

PS

Cisco Local Director IBM Network Dispatcher Microsoft SharePoint F5 Labs BIG/IP

29

Q: Compare Routing Policies for E[T]?

PS

RouterPS

PS

Web Server Farm

High variance job size

??

1. Random


3. Least-Work-Left Go to host with least total

work.

4. Size-Interval Splitting Jobs are split up by size

among hosts. (Central-Queue policies aren’t

possible for PS farms)

HighE[T]FCFS

LowE[T]FCFS

30

Q: Compare Routing Policies for E[T]? PS

RouterPS

PS

1. Random


3. Least-Work-Left Go to host with least total

work.

4. Size-Interval Splitting Jobs are split up by size

among hosts.

Answer:Same for E[T],but not great.Also, want to balance load!

Answer:Shortest-Queue isgreedier & better.

PS farm 1[ ]

1ii i

E T pp

M/G/1/PS 1[ ]

1E T

E[T]

JSQLWL

RAND SIZE

8 servers, = .9, C2=50

31

Prior Analysis of JSQ Routing

FCFS

FCFSJSQ

All prior JSQ analysis assumes FCFS servers

2-server:

[Kingman 61] , [Flatto, McKean 77], [Wessels, Adan, Zijm 91]

[Foschini, Salz 78], [Knessl, Makkowsky, Schuss, Tier 87]

[Conolly 84], [Rao, Posner 87], [Blanc 87], [Grassmann 80], [Muntz, Lui, Towsley 95]

[Cohen, Boxma 83]

>2-server approximations:

[Nelson, Philips, Sigmetrics 89][Nelson, Philips, Perf.Eval. 93]

[Lin, Raghavendra, TPDS 96]

32

[Nelson, Philips] Idea

FCFS

FCFSJSQ

Assume this is:

kMMn

//Assume this

is:

k: #servers

kn

n

nEnJobSizeE total] |queueshortest length [}allin totalPr{][

] [ TimeWaitingE

33

First Analysis of JSQ for PS [Gupta, Harchol-Balter, Sigman, Whitt, 06+]

JSQPoisson Process

PS

PS

PS

Near insensitivity to C2:PS server farm with General service PS server farm w/Exponential service FCFS server farm w/Exponential service

Single-queue equivalence:For PS server farm w/Exponential service,Multiserver system Single queue w/ contingent arrival rates

34

Summary so far

LESSONS LEARNED: Greedy routing policies, like JSQ, LWL are poor. To combat variability, need size-interval splitting. By isolating smalls, can achieve effects of smart single-server policies Load UN-balancing Don’t need to know size.

FCFS

FCFSRouter

Supercomputing

PS

RouterPS

PS

Web server farmLESSONS LEARNED: JSQ routing is good! Job size variability not a problem. Load Balancing

35

Outline

FCFS

FCFSRouter

I. Review of scheduling in single-serverM/GI/1

II. Supercomputing/Manufacturing


PSRouter

PS


SRPT

SRPT

SRPTRouter&

36

What is Optimal Routing/Scheduling?

Sched. policy

Router

Sched. policy

Sched. policy

Routingpolicy

Incomingjobs

2 Policy Decisions

Assume no restrictions: Jobs are fully preemptible. Can have central queue if want it, or not. Know job size (of course don’t know future jobs ...)

37

What is Optimal Routing/Scheduling?

SRPT

Central-Queue-SRPT

Question: Central-Queue-SRPT looks pretty good! Does it minimize E[T]?

Recall: minimizes E[T] on every sample path! [Schrage 67]

SRPT

38

Central-Queue-SRPT

SRPT

Answer: This does not minimize E[T] on every arrival sequence.

Bad Arrival Sequence:@time 0: 2 jobs size 29, 1 job size 210

@time 210: 2 jobs size 28, 1 job size 29

@time 210 + 29: 2 jobs size 27, 1 job size 28, etc.

2928

21029

2928

OPT

29

2102829

28

29

Central-Queue-SRPT

preempted

39

Central-Queue-SRPT

SRPT

Adversarial (Worst-Case) Guarantees:

THM: [Leonardi, Raz, STOC 97]: Central-Queue-SRPT is

competitive for E[T], and no online policy can improve upon this by more than constant factor.

serversjobs

sizesmallestsizebiggestO #

# ,minlog

Remarks: log(biggest/smallest) could be factor 7 in practice! Closest stochastic result analyzes only central-queue w/priorities: [Harchol-Balter, Wierman, Osogami, Scheller-Wolf, QUESTA 05]

40

What is Optimal Routing/Scheduling

with Immediate Dispatch?

Sched. policy

Router

Sched. policy

Sched. policy

Routingpolicy

Incomingjobs

2 Policy Decisions

Practical Assumption: jobs must be immediately dispatched! Jobs are fully preemptible within queue. Know job size.

41

What is Optimal Routing/Scheduling

when Immediately Dispatch?Claim: The optimal routing/sched.pair given immed. dispatchuses SRPT at the hosts.(Assuming an opt pair exists.)

PROOF: Let A: optimal routing/scheduling pair wrt E[T]. Suppose by contradiction: A does not use SRPT at the hosts. Let policy pair B mimic A with respect to A’s dispatching of jobs to hosts. I.e., policy B may be different from A, but sends the same jobs to the same hosts at the same times as A. But after the dispatching, B does SRPT scheduling at the hosts. Thus B improves upon A with respect to E[T]. Contradiction!

Router

SRPTImmediatelyDispatch Jobs

Incomingjobs SRPT

IMPACT: Claim narrow search to policies with SRPT at hosts.

42

In search of good Immediate Dispatch Routing

Q: What should immediate dispatch routing policy be, given SRPT sched. at hosts?

Router


Incomingjobs

SRPT

SRPT

43

Bad Arrival Sequence:@time 0: job of size 10 arrives.@time 0+: job of size 1000 arrives.@time 0++: job of size 10 arrives.@time 0+++: job of size 1 arrives. @time 1+++: job of size 1000 arrives.

JSQ/SRPT


SRPT

10

1000

10

1

1000

OPT

10

1000

10

11000

JSQ

In search of good Immediate Dispatch Routing

… why not obvious

44

Smart Immediate Dispatch Policy

Answer: IMD Algorithm due to [Avrahami,Azar 03]: Split jobs into size classes Assign each incoming job to server w/ fewest #jobs in that class

Router

SRPTImmediatelyDispatch

Incomingjobs

SRPT

SRPT

Remarks: IMD is competitive for E[T]. Immediate Dispatching is “as good as” Central-Queue-SRPT Similar policy proposed by [Wu,Down 06] for heavy-traffic setting.

jobsO smallestbiggest #,minlog

45

Some Key Points

FCFS

FCFSRouter

Supercomputing Web server farm model

PSRouter

PS

Towards Optimality …

SRPT

SRPT

SRPTRouter&

• Need Size-interval splitting to combat job size variability and enable good performance.

• Job size variability is not an issue.• Greedy, JSQ, performs well.

• Both these have similar worst-case E[T].• Almost exclusively worst-case analysis, so hard to compare with above results.• Need stochastic research here!

46

If you want to know more …

15-849 Performance Modeling15-849 Performance Modeling15-849 Performance Modeling15-849 Performance Modeling** Highly-recommended for CS theory, Math, TEPPER, and ACO doctoral students** Highly-recommended for CS theory, Math, TEPPER, and ACO doctoral students** Highly-recommended for CS theory, Math, TEPPER, and ACO doctoral students** Highly-recommended for CS theory, Math, TEPPER, and ACO doctoral students

Instructor: Mor Harchol-Balter ([email protected]) Instructor: Mor Harchol-Balter ([email protected])

www.cs.cmu.edu/~harchol/www.cs.cmu.edu/~harchol/Instructor: Mor Harchol-Balter ([email protected]) Instructor: Mor Harchol-Balter ([email protected])

www.cs.cmu.edu/~harchol/www.cs.cmu.edu/~harchol/

Queueing theory is an old area of mathematics which has recently become very hot. The goal of Queueing theory is an old area of mathematics which has recently become very hot. The goal of queueing theory has always been to improve the design/performance of systems, e.g. networks, servers, queueing theory has always been to improve the design/performance of systems, e.g. networks, servers, memory, disks, distributed systems, etc., by finding smarter schemes for allocating resources to jobs.memory, disks, distributed systems, etc., by finding smarter schemes for allocating resources to jobs.

In this class we will study the beautiful mathematical techniques used in queueing theory, including In this class we will study the beautiful mathematical techniques used in queueing theory, including stochastic analysis, discrete-time and continuous-time Markov chains, renewal theory, product-forms, stochastic analysis, discrete-time and continuous-time Markov chains, renewal theory, product-forms, transforms, supplementary random variables, fluid theory, scheduling theory, matrix-analytic methods, transforms, supplementary random variables, fluid theory, scheduling theory, matrix-analytic methods, and more. Throughout we will emphasize realistic workloads, in particular heavy-tailed workloads. and more. Throughout we will emphasize realistic workloads, in particular heavy-tailed workloads.

This course is packed with This course is packed with open problemsopen problems -- problems which if solved are not just interesting -- problems which if solved are not just interesting theoretically, but which have huge applicability to the design of computer systems today.theoretically, but which have huge applicability to the design of computer systems today.

Queueing theory is an old area of mathematics which has recently become very hot. The goal of Queueing theory is an old area of mathematics which has recently become very hot. The goal of queueing theory has always been to improve the design/performance of systems, e.g. networks, servers, queueing theory has always been to improve the design/performance of systems, e.g. networks, servers, memory, disks, distributed systems, etc., by finding smarter schemes for allocating resources to jobs.memory, disks, distributed systems, etc., by finding smarter schemes for allocating resources to jobs.

In this class we will study the beautiful mathematical techniques used in queueing theory, including In this class we will study the beautiful mathematical techniques used in queueing theory, including stochastic analysis, discrete-time and continuous-time Markov chains, renewal theory, product-forms, stochastic analysis, discrete-time and continuous-time Markov chains, renewal theory, product-forms, transforms, supplementary random variables, fluid theory, scheduling theory, matrix-analytic methods, transforms, supplementary random variables, fluid theory, scheduling theory, matrix-analytic methods, and more. Throughout we will emphasize realistic workloads, in particular heavy-tailed workloads. and more. Throughout we will emphasize realistic workloads, in particular heavy-tailed workloads.

This course is packed with This course is packed with open problemsopen problems -- problems which if solved are not just interesting -- problems which if solved are not just interesting theoretically, but which have huge applicability to the design of computer systems today.theoretically, but which have huge applicability to the design of computer systems today.

My class lectures are all available online.

47

N. Avrahami and Y. Azar, “Minimizing Total flow time and total completiontime with immediate dispatching.” SPAA 2003, pp. 11-18.

N. Bansal and M. Harchol-Balter, "Analysis of SRPT scheduling: Investigating unfairness," Proceedings of ACM Sigmetrics 2001.

P. Barford and M. Crovella, “Generating representative web workloads for network and server performance evaluation,” ACM Sigmetrics 1998, pp. 151-160.

J. Blanc, “A note on waiting times in systems with queues in parallel,” J. Appl. Prob., Vol. 24, 1987 pp 540-546.

S. Borst, O. Boxma, and P. Jelenkovic, “Reduced load equivalence and induced burstiness in GPS queues with long-tailed traffic flows,” Queueing Systems, Vol. 43, 2003, pp. 274-285.

S. Borst, O. Boxma, and M. van Uitert, “The asymptotic workload behavior of two coupled queues,” Queueing Systems, Vol. 43, 2003, pp. 81-102.

J.W. Cohen and O. Boxma, Boundary Value Problems in Queueing System Analysis, North Holland, 1983

B. Conolly, “The Autostrada queueing problem,” J. Appl. Prob.: Vol. 21., 1984, pp. 394-403.

References

48

M. Crovella and A. Bestavros, “Self-similarity in world wide web traffic: evidence and possible causes,” Proceedings of the 1996 ACM Sigmetrics International Conference on Measurement and Modeling of Computer Systems, May 1996, pp. 160-169.

D. Down and R. Wu, “Multi-layered round robin scheduling for parallel servers,” Queueing Systems: Theory and Applications, Vol. 53, No. 4, 2006, pp. 177-188.

G. Fayole and R. Iasnogorodski, “Two coupled processors: the reduction to a Riemann-Hilbert problem,” Zeitschrift fur Wahrscheinlichkeistheorie und vervandte Gebiete, vol. 47, 1979, pp. 325-351.

L. Flatto and H.P. McKean, “Two queues in parallel,” Communication on Pure and Applied Mathematics, Vol. 30, 1977, pp. 255-263.

R. Foley and D. McDonald, “Exact asymptotics of a queueing network with a cross-trained server,” Proceedings of INFORMS Annual Meeting, October 2003, pp. MD-062.

G. Foschini and J. Salz, “A basic dynamic routing problem and diffusion,” IEEE Transactions on Communications, Vol. Com-26, No. 3, March 1978.

References, cont.

49

P. Glynn, M. Harchol-Balter, K. Ramanan, “Heavy-traffic approach to optimizing size-interval task assignment,” Work in progress, 2006.

W. Grassmann, "Transient and steady state results for two parallel queues," Omega, vol. 8, 1980, pp. 105-112.

V. Gupta, M. Harchol-Balter, K. Sigman, and W. Whitt, “Analysis of join-the-shortest-queue policy for web server farms.” In submission, 2006.

M. Harchol-Balter and A. Downey. "Exploiting process lifetime distributions for dynamic load balancing," Proceedings of ACM Sigmetrics '96 Conference on Measurement and Modeling of Computer Systems , May 1996, pp. 13-24.

M. Harchol-Balter, M. Crovella, and C. Murta, "On choosing a task assignment policy for a distributed server system," Journal of Parallel and Distributed Computing , vol. 59, no. 2, Nov. 1999, pp. 204-228.

M. Harchol-Balter, C. Li, T. Osogami, and A. Scheller-Wolf, and M. Squillante, “Cycle stealing under immediate dispatch task assignment,” Proceedings of the Annual ACM Symposium on Parellel Algorithms and Architectures (SPAA), June 2003, pp. 274-285.

References, cont.

50

M. Harchol-Balter, B. Schroeder, N. Bansal, M. Agrawal. "Size-based scheduling to improve web performance." ACM Transactions on Computer Systems , Vol. 21, No. 2, May 2003, pp. 207-233.

M. Harchol-Balter and R.Vesilo, “Optimal cutoffs for size-interval task assignment,” Work in progress, 2006.

M. Harchol-Balter, A. Wierman, T. Osogami, and A. Scheller-Wolf, "Multi-server queueing systems with multiple priority classes," Queueing Systems: Theory and Applications (QUESTA), vol. 51, no. 3-4, 2005, pp. 331-360.

J. Kingman, “Two similar queues in parallel,” Biometrika, Vol. 48, 1961, pp. 1316-1323.

A. Konheim, I. Meilijson, and A. Melkman, “Processor-sharing of two parallel lines,” J. Appl. Prob., Vol. 18, 1981, pp. 952-956.

C. Knessl, B. Matkowsky, Z. Schuss, and C. Tier, “Two parallel M/G/1 queues where arrivals join the system with the smaller buffer content,” IEEE Transactions on Communications, Vol. Com-35, No. 11,1987, pp. 1153-1158.

S. Leonardi and D. Raz, “Approximating total flow time on parallel machines,”ACM Symposium on Theory of Computing (STOC), 1997.

References, cont.

51

H. Lin, and C. Raghavendra, “An approximate analysis of the join the shortest queue (JSQ) policy”, IEEE Transactions on Parallel and Distributed Systems, vol. 7, no. 3, March 1996.

J. Lui, R. Muntz, D. Towsley, “Bounding the mean response time of the minimum expected delay routing policy: an algorithmic approach,” IEEE Transactions on Computers, Vol. 44, No. 12, Dec 1995.

S. Muthukrishnan, R. Rajaraman, A. Shaheen, and J. Gehrke, “Online scheduling to minimize average stretch,” Proceedings of the 40th Annual Symposium on Foundations of Computer Science, October 1999, pp. 433.

R. Nelson and T. Philips, “An approximation to the response time for shortest queue routing,” ACM SIGMETRICS Performance Evaluation Review, Vol. 17 No. 1, May 1989, pp. 181-189.

R. Nelson and T. Philips, “An approximation for the mean response time for shortest queue routing with general interarrival and service times,” Performance Evaluation, Vol. 17 No. 2, March 1993 pp. 123-139.

References, cont.

52

References, cont.

T. Osogami, M. Harchol-Balter, and A. Scheller-Wolf, “Analysis of cycle stealing with switching cost,” Proceedings of the ACM Sigmetrics, June 2003, pp. 184-195.

T. Osogami, M. Harchol-Balter, and A. Scheller-Wolf. "Analysis of cycle stealing with switching times and thresholds" Performance Evaluation, Vol. 61, No. 4, 2005, pp. 347-369.

B. Rao and M. Posner, “Algorithmic and approximate analysis of the shorter queue,” Model Naval Research Logistics, Vol. 34, 1987, pp. 381-398.

R. Righter and J. Shanthikumar, “Scheduling multiclass single server queueing systems to stochastically maximize the number of successful departures," Probability in the Engineering and Informational Sciences, Vol. 3, 1989, pp. 323-333.

L.E. Schrage, “A proof of the optimality of the shortest processing remaining time discipline,” Operations Research, Vol. 16, 1968, pp. 678-690.

B. Schroeder and M. Harchol-Balter, "Evaluation of task assignment policies for supercomputing servers: The case for load unbalancing and fairness," 9th IEEE Symposium on High Performance Distributed Computing (HPDC '00) , August 2000.

53

References, cont.

B. Schroeder, A. Wierman, and M. Harchol-Balter. "Closed versus open system models: A cautionary tale,” Proceedings of NSDI , 2006.

A. Shaikh, J. Rexford, and K. Shin, “Load-sensitive routing of long-lived IP flows,” Proceedings of SIGCOMM, September, 1999.

J. Wessels, I. Adan, and W. Zijm, “Analysis of the asymmetric shortest queue problem,” Queueing Systems, Vol. 8, 1991, pp. 1-58.

A. Wierman and M. Harchol-Balter. "Classifying scheduling policies with respect to higher moments of conditional response time." Proceedings of ACM Sigmetrics 2005 Conference on Measurement and Modeling of Computer Systems.

A. Wierman and M. Harchol-Balter. "Nearly insensitive bounds on SMART scheduling." Proceedings of ACM Sigmetrics 2005 Conference on Measurement and Modeling of Computer Systems.

A. Wierman and M. Harchol-Balter, "Classifying scheduling policies with respect to unfairness in an M/GI/1," Proceedings of ACM Sigmetrics 2003 Conference on Measurement and Modeling of Computer Systems , June 2003.

1 scheduling in server farms mor harchol-balter computer science dept carnegie mellon university...

Documents