An Adaptive Data Placement Architecture in Multicloud Environments

Algorithm 3

Algorithm DPQ: data placement algorithm based on Q-learning.

	Input: The state matrix, , the transfer matrix, .
	Output: The optimal data placement solution during ,
(1)	Initialize parameters of algorithm including learning rate , discount
(2)	Initialize the tabular with zero;
(3)	Initialize the start state;
(4)	Initialize data placement sequences ;
(5)	for to do
(6)	for to do
(7)	Choose feasible data placement solutions through equation
(8)	Choose a data placement solution from through -greedy function;
(9)	Append in ;
(10)	Obtain the next state , reward , and the next state’s
(11)	;
(12)	;
(13)	Take decay;
(14)	end for
(15)	end for
(16)	Find the sequence with lowest cost in ;
(17)	return;