Non-preemptive Coflow Scheduling and Routing Ruozhou Yu , Guoliang - - PowerPoint PPT Presentation

▶

Oct 20, 2022 49 likes •272 views

IEEE Globecom 2016 SAC-ANS 3 Non-preemptive Coflow Scheduling and Routing Ruozhou Yu , Guoliang Xue, and Xiang Zhang Arizona State University Jian Tang Syracuse University 1/22 Outline q Introduction and Motivation q System Model and

SLIDE 1

Non-preemptive Coflow Scheduling and Routing

Ruozhou Yu, Guoliang Xue, and Xiang Zhang Arizona State University Jian Tang Syracuse University

1/22 IEEE Globecom 2016 SAC-ANS 3

SLIDE 2

Outline

q Introduction and Motivation q System Model and Algorithm Design q Performance Evaluation q Conclusion

2/22

SLIDE 3

Flows and Coflows

q

3/22 OMG where’s my last piece??? Need it now!!! Data piece 1 Data piece 2 Data piece 3 Data piece 4 Data piece 5 Data piece 6

Cloud Task

SLIDE 4

Flows and Coflows

q Traditional network scheduling/routing solution

v Scheduling/Routing regarding individual flows

vGeneral flow: a subset of packet header fields

v Fails to account for application-level performance metrics

vFlow completion time vs. task completion time 4/22

SLIDE 5

Flows and Coflows

5/22 You faster, you go ahead J Oh you’re so nice!!

Task 1 Task 2

SLIDE 6

Flows and Coflows

q Application-aware scheduling/routing: coflows

v Flows grouped by application/task

vA coflow finishes when all its component flows finish

v Advantages:

vCaptures application-level requirement vEstablishes fairness in application-level

v Want to do it in a centralized way

vNot to leak app privacy to other apps vOr to prevent apps from selfishly congest the network 6/22

SLIDE 7

(Non-)Preemptive Scheduling

q Existing coflow scheduling/routing allows preemption!

v Pause for the shorter ones! v Advantages:

vBetter performance and network utilization in theory

v Disadvantages:

vLarge overhead for flow switching: performance issue for short flows

q Switching delays q Switch computations

vNo ready support in commodity hardware

q Standardization on-going: IEEE 802.1Qbu q A long way before commercial-ready

q Our stand: non-preemptive scheduling + routing of coflows

7/22

SLIDE 8

Summary of Problem

8/22

BOSS

Now, you go first, this way! You next, that way! You this way, free to go! Sorry, there’s no place. You fired!

Task 1 Task 2 Task 3 Task 4

SLIDE 9

Contributions

q A first (preliminary) study for Non-preemptive Coflow Scheduling and Routing (NCSR) q An offline scheduling framework: Shortest-Coflow First q A multi-path routing algorithm q A single-path routing algorithm q Performance evaluations

9/22

SLIDE 10

Outline

q Introduction and Motivation q System Model and Algorithm Design q Performance Evaluation q Conclusion

10/22

SLIDE 11

System Model

11/22

q Network: G = (V, E) q Coflow requests: S = {C1, …, Cm}

v Each request: Ci = {Fi,1, …, Fi,ni}

vFi,j = (si,j, ti,j, di,j): source, destination, flow size (demand, in

bytes)

q Bandwidth allocation vBp

i,j(t): bandwidth allocation on path p of flow i, j, at time t

vBi,j(t) = sum of bandwidth over all paths at time t

SLIDE 12

System Model

q Flow/coflow completion time

v Flow completion time (FCT): v Coflow completion time (CCT): max. FCT of its component flows v Objective: minimize total CCT

12/22

SLIDE 13

Shortest-Coflow First Scheduler

q For each coflow:

v Compute per-coflow completion time (CCT)

vIf multi-path enabled, compute using multi-path routing vOtherwise, use single-path routing

q Schedule coflows in ascending order of CCT

13/22

SLIDE 14

CCT with Multi-path Routing

q Non-linear programming formulation

v Sharing among flows within the coflow v CCT as the maximum FCT of component flows

q Linearization: let fi = 1 / Ti

14/22

SLIDE 15

CCT with Single-path Routing

q Additional integer variables to the Multi-path Routing model vxe

i,j: link selection for single-path routing

q Linear relaxation and deterministic rounding

v Relax xe

i,j to take continuous values, and solve linear program;

v For each flow, find path with maximum minimum x values, and assign; v Re-solve program to obtain bandwidth allocation with fixed path assignments

15/22

SLIDE 16

Outline

q Introduction and Motivation q System Model and Algorithm Design q Performance Evaluation q Conclusion

16/22

SLIDE 17

Simulation Setups

q Waxman random graphs

v 50 nodes v Alpha=0.15, beta=0.2 v Link capacities: [10, 100] Mbps

q Coflows

v 25 requests v 1 to 10 flows per request v Flow sizes: [10, 100] Mbps

q Comparison:

v sSCF, mSCF: single-path and multi-path SCF algorithm (proposed) v sRT, mRT: single-path and multi-path Routing-only algorithm (baseline) v sSFF, mSFF: single-path and multi-path Shortest-Flow First (baseline) 17/22

SLIDE 18

Simulation Results: Average CCT

18/22

SLIDE 19

Simulation Results: Running Time

19/22

SLIDE 20

Outline

q Introduction and Motivation q System Model and Algorithm Design q Performance Evaluation q Conclusion

20/22

SLIDE 21

Conclusions

q A first step study on NCSR

v Offline optimization model v SCF scheduler for scheduling v Multi-path and single-path routing algorithms

q Experiment results

v Scheduling more effective than routing: when network congested v Application-awareness brings great advantage

q Future work

v Enable better sharing/work conservation of resources

vRemove the non-sharing rule of coflows 21/22

SLIDE 22

THANK YOU VERY MUCH!

Q&A?

22/22