Cost-effective data reliability assurance for data maintenance - Reliability Assurance of Big Data in the Cloud - page 47

Figure 6.1 PRCR architecture

As shown in Figure 6.1 , there are two major parts of PRCR, which are the user
interface and the PRCR node. Each of the components is deployed onto a Cloud
compute instance. For providing different data reliability assurances and managing
the huge number of Cloud data files, the entire PRCR mechanism could be composed
of one user interface and multiple PRCR nodes. Specifically, each PRCR node pro-
actively checks data files with a certain frequency, so that different data reliability
assurances can be provided by different PRCR nodes, which correspond to RA (1)
k

as demonstrated in Chapter 5 . The number of data files that a PRCR node manages is
referred to as the capacity of the PRCR node. As the maximum capacity of each PRCR
node is limited, each PRCR node could only manage a certain number of data files.
Each PRCR node works independently of one another, so that it can be easily created
and removed according to the number of data files that needs to be managed and the
data reliability assurance that needs to be provided.

6.2.1 User interface
The user interface is the component of PRCR responsible for determining the mini-
mum replica number, creating replicas by coordinating with the data creation service
(for storage with two replicas), creating and distributing metadata of data files, and
so forth.

• First, when the original replica of a data file is created (generated or uploaded) in the Cloud,
based on the minimum replication approach presented in Chapter 5 , the user interface deter-
mines the minimum number of replicas (i.e., one or two replicas).
• Second, if a data file needs to be stored with two replicas, the user interface calls Cloud
service to create the second replica for the data file, where the data transfer approach that we
propose (see Chapter 7 ) can be applied.
• Third, if a data file is stored with two replicas, the metadata of the data file are created and
distributed to the corresponding PRCR node.