Minimum replication for meeting the data reliability requirement - Reliability Assurance of Big Data in the Cloud - page 37

Minimum replication for meeting
the data reliability requirement

5

In this chapter we present the approach for calculating the minimum replication for
meeting the data reliability requirement. Essentially, based on the generic data re-
liability model, this approach provides a practical and efficient calculation method
with given parameters. When the data reliability requirement is determined and the
expected storage duration is provided, this approach could quickly calculate the mini-
mum replicas that are needed as well as predict the longest storage duration of the data
file for meeting the data reliability requirement. These outcomes of our approach are
the key for our data reliability assurance solution, based on which the whole series of
approaches during different data life cycle stages can be conducted. In addition, as
a direct consequence, the minimum replication can also act as a benchmark, which
can be used for evaluating cost-effectiveness for data reliability assurance of various
replication-based data storage approaches.
The structure of this chapter is organized as follows. In Section 5.1 , details of
the minimum replication calculation approach are presented. In Section 5.2 , we dis-
cuss about the application of the minimum replication benchmark for evaluation of
replication-based data storage approaches. In Section 5.3 , the outcomes of the evalu-
ation for the minimum replication calculation approach are briefly presented. Finally,
in Section 5.4 , we summarize the works presented in this chapter.

5.1 The minimum replication calculation approach

As mentioned previously, our minimum replication calculation approach has two pur-
poses. First, it determines the minimum replica number for ensuring the data reli-
ability requirement. Second, given a certain data reliability requirement, it predicts
the longest storage duration of the data file while the data reliability requirement is
met. By solving our generic data reliability model presented in Chapter 4 , the longest
storage duration of Cloud data files with any number of replicas can be predicted,
however, considering that no more than two replicas for each data file are needed in
our data storage solution. Therefore, in this section, we only present the investigations
conducted for the Cloud data files stored with a single replica or two replicas.

5.1.1 Minimum replication calculation formulas
In a commercial storage system such as that of the Cloud, “data reliability” has two
aspects of meaning, which are the data reliability requirement RR ( t ) and the data reli-
ability assurance RA ( t ). RR ( t ) indicates the data reliability that storage users wish to
achieve within the storage duration of t , while RA ( t ) indicates the data reliability that