Вы находитесь на странице: 1из 5

Scientific Journal of Information Engineering

Xin Xi Gong Cheng Qi Kan

116023

Query Schedule Algorithm Based on Deadline and Value over Data


Stream System
Lijie Zhang, Haiyan Liu
Software College Dalian University of Foreign Languages, 116044, China
Email: lijie8111@126.com
Abstract: Dynamic stream rate of continuous queries, real-time deadline and important value in high critical data stream application is
considered in this paper. Firstly, the relative deadline and value is calculated real-time. Secondly, a priority table based on the relative
deadline and value is created and filled which decides the query scheduling order. At last, based on this priority table, a value and deadline
jointly based query scheduling algorithm in different stream radio is proposed. The experimental results indicate that the algorithm raises
the hit value radio (HVR) and the success ratio continuous query scheduling.
Key words: Data stream; Deadline; Value; Priority; Query scheduling

earliest deadline
first EDFleast slack first LSFhighest value first
HVF[1]

[2]
QoS [3]
[4]

90 | SJIE Jun. 2012 V ol. 2 No.3 PP.90-94 www.sjie.org 2011 American V -King Scientific Publishing, LTD

Xin Xi Gong Cheng Qi Kan

Scientific Journal of Information Engineering

A
R

t
di
vi
pi

tuples/second

e=e +c

a di

si

si

2
1

2.1 Di
a s i di di -ai a

Di =di -ai

(1)

2.2 Vi
s i
Vi t s i vi di a+r*di r>=1 0
r>1 Vi

Vi t vi ,

0 t di

rdi (r 1)a t (r 1) Di vi ,
0,

d i t a rdi

(2)

a rdi t

3 -

3.1 -

SJIE Jun. 2012 V ol.2 No.3 PP.90-94 www.sjie.org 2011 American V -King Scientific Publishing, LTD | 91

Xin Xi Gong Cheng Qi Kan

Scientific Journal of Information Engineering

V1 V4 D1 D4
4
3
V1 1

0 Vi 1 vi 4

V2 2

V
V
V

V3 3
V4 4

4 Vi 1Vi 4 2

n
1 i

n
1 i

4 2 Vi 1Vi 4 3

4 3 Vi 1Vi

(3)

1 i

4
D1 1

0 Di 1 Di 4

D2 2

D
D
D

D3 3
D4 4

4 2 Di 1 Di 4 3

4 3 Di 1 Di

4 Di 1 Di 4 2

(4)

s i

3.2
r p
1 1/r > e

5 P
6
Pi Vi Di

(5)

Pi Vi D j

(6)

2 1/r <= e

5 P 7

Pi (Vi Di 1) (Vi Di 2) 2 D j

(7)

p
2

StreamScheduling SS
d,v,r,e,a,t

92 | SJIE Jun. 2012 V ol. 2 No.3 PP.90-94 www.sjie.org 2011 American V -King Scientific Publishing, LTD

Xin Xi Gong Cheng Qi Kan

Scientific Journal of Information Engineering

V1

V2

V3

V4

V1

Value

V2

V3

V4

D1

D1

D2

0.5

1.5

D2

12

D3

0.33

0.67

1.33

D3

13

18

D4

0.25

0.5

0.75

D4

10

14

19

25

Value

Deadline

Deadline

init(); */
1.

Di = computeDeadline(d,a); 1

2.

Vi =computeValue(v,r,d,a,t); 2

3.

createTable(Di ,Vi ); 3,4

4.

if(1/r>=e){

5.

confirmCell(Di ,Vi ) 5,6

6.

traversalTable(Di ,Vi ); } 1

7.

if(1/r<e) {

8.

confirmCell(Di ,Vi ) 5,7

9.

traversalTable(Di ,Vi ); } 2

10. updateQueryPlan();

5
Microsoft WinXP Professional64 AMD 3000+ 1G
HVR x y
HVR HVR
EDF HVF SS
3

EDF

HVF

SS

HVR

100
90
80
70
60
50
40
30
20
10
0
0.5

1.5

2.5

3.5

SJIE Jun. 2012 V ol.2 No.3 PP.90-94 www.sjie.org 2011 American V -King Scientific Publishing, LTD | 93

Xin Xi Gong Cheng Qi Kan

Scientific Journal of Information Engineering

3
SS HVR 80%

[1]

Buyya R, M urshed M , Abramson D, et al. Scheduling paramenter sweep applications on global grids: A deadline and buget constrained
cost-time optimization algorithm[J]. Software: Paractice and Experience (SPE), 2005, 35(5): 491-512.

[2]

S. Viglas,J. F. Naughton: Rate-Based Optimization for Streaming Information Sources[J]. SIGM OD Conference, 2002.

[3]

D J Abadi, D Carney, et al. Aurora: a new model and architecture for data stream management[J], The VLDB Journal, 2003, 12(2): 120.

[4]

D. Carney, U. etintemel, A. Rasin, et al. Operator Scheduling in a Data Stream M anager[J]. Department of Computer Science, Brown
University, Department of Computer Science, Brandeis University, Laboratory for Computer Science & Department of EECS,
M assachusetts Institute of Technology . 2003, 29: 838-849.

1981-2003-2006
2006-
Email: lijie8111@126.com

94 | SJIE Jun. 2012 V ol. 2 No.3 PP.90-94 www.sjie.org 2011 American V -King Scientific Publishing, LTD

Вам также может понравиться