Generic data reliability model in the cloud - Reliability Assurance of Big Data in the Cloud - page 31

Generic data reliability model
in the cloud

4

In this chapter, we present our replication-based data reliability model in detail. In
this model, the relationships among reliability of Cloud data, storage duration, storage
device failure pattern, and replication level are well described. By conducting further
investigation on properties of the data reliability model and describing detailed deri-
vation of the generic data reliability model step by step, this chapter clearly presents
the solid theoretical foundation of our research, which can also be easily understood
by the readers.
The structure of this chapter is organized as follows. In Section 4.1 , further analy-
sis on properties of the data reliability model is conducted, in which several aspects
of the model, such as properties of the model and the disk failure rate pattern are
determined. In Section 4.2 , the detailed derivation of the generic data reliability
model is presented step by step, in which data reliability with static disk failure
rate, data reliability of a single replica with variable disk failure rate, and finally the
generic data reliability model for multiple replicas with variable disk failure rate are
described respectively. Finally, in Section 4.3 , we summarize the works presented
in this chapter.

4.1 Properties of the data reliability model

In order to design a data reliability model with all the requirements listed in Chap-
ters 1 and 3 being met, further analyses on more detailed properties of the data reli-
ability model need to be conducted. Therefore prior to the presentation of our generic
data reliability model, in this section we determine the reliability metrics that we use,
the model type that we use for the model design as well as the failure rate pattern of
storage devices that are applied for describing Cloud data reliability and storage de-
vices reliability respectively, and explain the reason why these specific properties are
selected.

4.1.1 Reliability metrics
As mentioned in Section 2.1 , there are two fundamental disk reliability metrics that
are currently used for describing the permanent disk failure rates, which are the mean
time to failure ( MTTF ) and annualized failure rate ( AFR ). In this topic, we apply the
AFR as the disk reliability metric to our research because of the following two reasons.

• First, AFR is easier to be understood by nonexpert readers. The represen-
tation of MTTF is by time, which is calculated according to the equation
= MTTF TestDiskNumber TestHoursDiskFailures ⋅ / . For example, a disk manufacturer