Database Reference
In-Depth Information
Fig. 8.6
Crossover and mutation operators
function must penalize solutions that do not meet the QoS constraints and drive
the evolution towards satisfaction. The distance from constraint satisfaction for a
solution c is defined as:
0QC i .c/ 0
1QC i .c/ > 0
D.c/ D ˙ i D 1 QC i .c/ e i
w eight i ;e i
D
(8.1)
where weight i indicates the weight of the QoS constraint. Notice that this distance
function for constraints include both local and global constraints specified. The
fitness function for a chromosome c is then defined as follows:
F.c/ D ˙ i D 1 w i
Q i .c/ C w eight p D.c/
(8.2)
w i are the weights for each QoS attribute. weight p is the penalty factor . Several
features are highlighted when calculating the fitness function based on the match
string and the scheduling string:
1. Services are executed exactly in the order specified by the scheduling string. For
example, Fig. 8.7 a shows a scheduling string for a composition. Assume there
are two different match strings for this ss.(a)ms 1 :LetS 1 and S 2 be assigned to
the same VM v m 1 , and S 3 be assigned to another VM v m 2 . In this chromosome,
because S 1 is to be executed before S 2 , data 1 is available before data 2 . Thus,
data 1 will be transferred to S 3 before data 2 .And data 1 will be stored in S 3 's
database service till data 2 has been transferred to S 3 .(b)ms 2 : Let the three
services S 1 , S 2 , and S 3 be assigned to three different VMs v m 1 , v m 2 and v m 3 .
Search WWH ::




Custom Search