1. A. Rasoolzadegan R. Alesheykh A. Abdollahzadeh
A New Approach for Event Triggering Probability Estimation in Active Database Systems to Rule Scheduling Improvement 2nd International Conference on Information & Communication Technologies: from Theory to Applications (ICTTA)
AUT
CE & IT
A. Rasoolzadegan
R. Alesheykh
A. Abdollahzadeh
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

2. Passive versus Active Databases
Passive databases only execute explicit queries and transactions.
Semantics of the domain are expressed in code associated with application programs.
Constraints and semantics distributed across many application programs.
Database may be updated without “going through” an application program, thereby making the database inconsistent with the rules governing the data.
Users and application programs need to periodically poll the database to verify if anticipated conditions are valid.
Could become a performance bottleneck
The applications and knowledge-based systems interfacing with the database may be out of sync with one another.
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

3. Active Databases
An active database is capable of monitoring the database state to
Alert applications programs, expert systems, and more general knowledge-based systems,
Update the database through rule-generated updates.
Shared application semantics may be embedded within the database itself.
Research has focused on integrating AI production systems with the database.
Issues related to expressing the rules, their execution models, and rule optimization.
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

4. Rules in Active Databases
Rules provide a uniform and powerful mechanism to express:
Integrity constraints and their enforcement
Assertions about the database
Triggers and alerts
Corporate Policy Specification
Configuration Management Policies
Update semantics for shared objects with alerts to notify of changes to relevant objects
Performance monitoring for 1) load balancing, 2) query optimization, and database reorganization
Usage monitoring for knowledge discovery regarding user profiles and preferences.
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

5. Rule Processing Unit 5 1 2 4 3 3 External Events Temporal Events
Application
……………….
raise-event(e1)
………………..
External Events
Event Detection
Internal Events
Rules Execution 5
Temporal Events 1
Rule Processing Unit 2 4
Condition Evaluation 3 3
Rule selection
Active Rules Set
Data Flow
Control Flow
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

6. coupling modes in Active Database
Immediate
Deferred
Independent
Child Transaction
Parent Transaction T t0 t1 t2 t3 t4 t5 t6 t7 t8 t9
t10
S(T)
F(T)
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

7. Ex-SJF Rule Scheduling Method
Based on SJF Algorithm
SJF algorithm is not useful for rule scheduling in ADBS due to active work load nature of it
there is defined preprocess for preparing rule base to use the SJF algorithm for rule scheduling in Ex-SJF
Rule Execution Tree
Virtual Rule
Exact Rule Execution Time Calculation
Two version of Ex-SJF:
Ex-SJFEXA
Ex-SJFPRO
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

8. Preparing of Rules Base in Ex-SJF
Rules Definition
Rules Execution Tree construction
Virtual Rules Construction
Rule Execution Time Calculation in Ex-SJFEXA or Ex-SJFPRO
Rules Base
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

9. Rule Execution Tree in Run Time
s(T) t1 t0
t20
t19
t21
t23
t25
t28
t26
t30
c(T)
f(T) t6 t7 t9
t10
t14
t17
t18
t22
t24
t27
t29 T
a(T) T
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

10. Rule Execution Tree Construction before Running
imm
def
R15
R12
R16
def
imm
R13
R10 R9 R5
R14
imm
def
Virtual Rules Extraction
Rule Execution Tree Construction R8 R4
imm
R13 R9
R14
def
imm R1
R13
R10 R8 R9 R7 R6 R5 R4
R11
R12 R3 R2
R14
imm
ind
def
R15
R16
R15 R1
R13
R10 R8 R9 R7 R6 R5 R4
R11
R12 R3 R2
R14
R16
imm
def
Omitting of Independent Rule Ledges
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

11. Ex-SJFEXA: Extended Shortest Job FirstEXACTLY
R1,X1
R8,X8
R9,X9
R7,X7
R6,X6
R5.X5
R4,X4
R3,X3
R2,X2
R13,X13
R14,X14
R12,X12
R11,X11
R10,X10
R19,X19
R20,X20
R18,X18
R17,X17
R16,X16
R25,X25
R26,X26
R24,X24
R23,X23
R22,X22
R21,X21
R15,X15 …
1 ………………………
2 ………………...
I ………………...
I+1 ……...
n ……………
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

12. The post order traversal of Rule Execution Tree in Ex-SJFEXA
def
imm
imm
imm
R3,3
R4,1
R4,3
R5,16
R5,2
R6,5
R6,2
R7,11
R7,5
def
imm
def
imm
def
R8,2
R9,4
R9,7
R10,7
R11,3
R12,1
R12,6
def
def
imm
imm
R13,2
R14,1
R15,3
R16,2
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

13. Ex-SJFPRO: Extended Shortest Job FirstPROBABILISTIC
R1,X1,P1
R8,X8,P8
R9,X9,P9
R7,X7,P7
R6,X6,P6
R5.X5,P5
R4,X4,P4
R3,X3,P3
R2,X2,P2 …
1 ………………………
2 ………………...
I ………………...
I+1 ……...
n ……………
R13,X13,P13
R14,X14,P14
R12,X12,P12
R11,X11,P11
R10.X10,P10
R18,X18,P18
R19,X19,P19
R17,X17,P17
R16,X16,P16
R15.X15,P15
R20,X20,P20
R24,X24,P24
R25,X25,P25
R23,X23,P23
R22,X22,P22
R21.X21,P21
(4-2)
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

14. The post order traversal of Rule Execution Tree in Ex-SJFPRO
def
imm
imm
imm
R3,3,0.5
R4,1,0.6
R4,1.2,0.6
R5,2,0.8
R5,7.6,0.8
R6,3.2,0.1
R6,2,0.1
R7,5,0.9
R7,6.6,0.9
def
imm
imm
def
def
R8,2,0.1
R9,4,0.2
R9,5.3,0.2
R10,7,0.6
R11,3,0.4
R12,1,0.5
R12,3.2,0.5
def
def
imm
imm
R13,2,0.4
R14,1,0.5
R15,3,0.6
R16,2,0.2
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

15. Evaluation Parameters
N = Number of Executed Rules
ART= Average Response Time
RTSV= Response Time Standard Variance
UCPU = CPU Utilization
TOPT = Time Overhead Per Transaction
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

16. Benchmarking Deferred mode Immediate mode Composite mode
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

17. Time Overhead per Transaction
Results of simulation of available rule scheduling approaches in deferred mode
CPU Utilization
Time Overhead per Transaction
Throughput
Response
Time Variance
Average Response Time
Evaluation
Parameters
Approaches 3 1 2
Random
Static Priority
FCFS 5
EDFPD
EDFDIV 4
EDFSL
Ex-SJFEXA
Ex-SJFPRO
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

18. Results of simulation of available rule scheduling approaches in immediate mode
CPU Utilization
Time Overhead per Transaction
Throughput
Response Time Variance
Average Response Time
Evaluation
Parameters
Approaches 3 2
Random
Static Priority 1
FCFS
EDFPD
EDFDIV
EDFSL
Ex-SJFEXA 4
Ex-SJFPRO
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

19. Results of simulation of available rule scheduling approaches in composite mode
CPU Utilization
Time Overhead per Transaction
Throughput
Response Time Variance
Average Response Time
Evaluation
Parameters
Approaches 1 3 4
Random 2
Static Priority
FCFS
EDFPD
EDFDIV 5
EDFSL
Ex-SJFEXA
Ex-SJFPRO
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

20. Ex-SJFPRO-V.1.8 Scheduling Approach
In this Method at first, execution time of each rule is calculated in the Ex-SJFPRO approach
Then in every time which an activated rule such as R1 is selected according to scheduling approach from activated rules list for condition evaluation, correctness probability of each logical statement of condition part of R1, i.e. P(R,LSi) is repeatedly calculated and stored based on :
This process is repeated for each logical statement of condition part of the corresponding rule until the changes rate of correctness probability of that logical statement, achieves to a desired value (e.g., and in general mode )
At this time, new value is replaced with primary default value (i.e. ½)
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

21. Percentage of rule scheduling optimization in Ex-SJFPRO-V. 1
Percentage of rule scheduling optimization in Ex-SJFPRO-V.1.8 in compare with Ex-SJFPRO
Throughput
Response Time
Variance
Average
Evaluation
parameters
Mode 9%
12.6% 8%
Immediate
14.3%
33.6%
18.8%
Deferred
12%
21.4%
14.2%
Composite
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

22. References
C. J. Date, “An Introduction to Database Systems”, 8th Edition, Addison-Wesley, 2003.
E. Bertino, B. Catania and G. P. Zarri, "Intelligent Database Systems", Addison-Wesley, 2001.
S. Potaminsto and M. Stonebraker, "The POSTGRES Rule System", in Active Database Systems: Triggers and Rules for Advanced Systems, Morgann Kaufmann Publishers, Sanfrancisco, CA, 1996.
H. Fritshi and Z. Flaach, “A Component Framework to Construct Active Database Management Systems”, PhD Thesis, CS Department, University of Zurich, 2002.
Vadua, "Rule Development for active database", PhD Thesis, CS Department, University of Zurich, 1999.
Vaduva, S. Gatziu and K. R. Dittrich, "Investigating Termination in Active Database Systems with Expressive Rule Languages". In Proceedings of the 3rd International Workshop on Rules In Database Systems, pp , Skovde(Sweden), 1997.
R. M. Sivasankaran, J. A. Stankovic, D. Towsley, B. Purimetla and K. Ramamritham, "Priority Assignment in Real-Time Active Databases", The International Journal on Very Large Data Bases, Vol. 5, No. 1, January 1996.
S. Ceri, C. Gennaro, S. Paraboschi and G. Serazzi, “Effective Scheduling of Detached Rules in Active Database”, IEEE Transaction on Knowledge and Data Engineering, Vol. 15, No.1, January/February 2003.
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory

23. References
R. Alesheykh, A. Abdollahzadeh, "Evaluation of Shortest Job First Approach for Rule Scheduling in Active Database Systems", in the Proceedings of the 9th IASTED International Conference on Artificial Intelligence & Soft Computing (ASC'05), Benidorm, Spain, September 2005.
S. Chakravarthy, "Architectures and monitoring techniques for active databases: An evaluation". In Technical Report TR , University of Florida, Gainesville, 1992.
J. Stankovic, M. Spuri, M.D. Natale and G.C. Buttazo, ”Implications of Classical Scheduling Results for Real-Time Systems”, IEEE Computer Society Press, Los Alamitos, CA, 1995.
Geppert, S. Gatziu, K. R. Dittrich, H. Fritschi, and A. Vaduva, "Architecture and implementation of the active object-oriented database management system SAMOS", Technical Report 95.29, CS Department, University of Zurich, 1995.
H. Theodore, "A survey of Active Database Systems", available in active.ps, April 1997.
R. Alesheykh, A. Abdollahzadeh, "Evaluation and Comparison of Rule Scheduling Approaches in Active Database Systems", in the Proceedings of the 2nd IASTED international Multi-Conference on Automation, Control, and Information Technology (ACIT’05), Russia, June 2005.
Department of Computer Engineering and Information Technology - Amirkabir University of Technology - Intelligent Systems Laboratory