Cost-effective data transfer for data creation and data recovery - Reliability Assurance of Big Data in the Cloud - page 65

Cost-effective data transfer for
data creation and data recovery

7

Data transfer activities are intensively involved in the replication-based Cloud data
storage process especially for data creation and data recovery. For cost-effective
big data storage in the Cloud while meeting the data reliability throughout the
Cloud data life cycle, the energy consumption of data transfer must be taken into
account. Based on previous analysis in Sections 2.3 and 3.2 on the features of the
Cloud network and data transfer in the Cloud, in this chapter we present our data
transfer solution that cooperates with PRCR for maintaining the huge number of
Cloud data files.
In this chapter, we present in detail our novel energy-efficient data transfer strat-
egy called Link Rate Controlled Data Transfer (LRCDT). By scheduling bandwidth
in a link rate controlled fashion, LRCDT aims to reduce the energy consumption
by “lazy” data transfer that does not require data to be transferred at the maximum
speed, so that the cost incurred for both data creation and data recovery processes
can be reduced.
The structure of this chapter is organized as follows. In Section 7.1 , we explain how
the data transfer deadline is determined for data creation and data recovery processes
respectively. In Section 7.2 , we describe our network model for Cloud networks with
bandwidth reservation. In Section 7.3 , we describe the energy consumption model
of network devices. In Section 7.4 , the data transfer strategy LRCDT is presented in
detail. In Section 7.5 , evaluation for LRCDT is conducted from the aspects of energy
consumption and task completion time respectively. Finally, in Section 7.6 , we sum-
marize the works presented in this chapter.
This section is mainly based on our work [87] .

7.1 Determining the deadline for data creation

and data recovery

As discussed in Chapter 5 , there is a time window for the system to increase the data
redundancy level when a data file is first created or needs to be recovered. To conduct
the data transfer activity for data creation or data recovery, in this section we discuss
how to determine the time window, that is, the data transfer deadline. Although the
process of data creation and data recovery are similar, due to different start times for
conducting the data transfer tasks, the deadline for data creation and data recovery are
determined based on different equations.

• When the Cloud data file is first created, if it needs to be stored with two replicas, there is
a deadline for generating the second replica. For the data files that are first created in the
Cloud, as the original replicas are newly generated and uploaded, to never get lost before