Consistency Management in Cloud Storage Systems - Large Scale and Big Data: Processing and Management - page 332

intuitive goals of weak ordering. Numerous variation models have been proposed
since. Bisiani et al. [22] proposed an implementation of weak consistency on dis-
tributed memory systems. Timestamps are used to achieve a weak ordering of the
operations. A synchronization operation is completed only after all previous opera-
tions in the systems reach a completion state. Various weaker consistency models
derived from the weak ordering. The following client-side models are weak consis-
tency models, but provide further guarantees to the client.
10.3.2.2 Client-Side Guarantees
10.3.2.2.1 Read-Your-Writes
This model guarantees that a process that commits an update will always be able to
see its updated value with the read operation and not an older one. This might be an
important consistency property to provide with weakly ordered systems for a large
class of applications. As will be seen further in this section, this is a special case of
causal consistency.
10.3.2.2.2 Session Consistency
Read-your-writes consistency is guaranteed in the context of a session (which is a
sequence of accesses to data, usually with an explicit beginning and ending). As
long as the users access data during the same session, they are guaranteed to access
their latest updates. However, the read-your-writes property is not guaranteed to be
spanned over different sessions.
10.3.2.2.3 Monotonic Reads
A process should never read a data item value older than what it has read before.
This consistency guarantees that a process's successive read returns always the same
value or a more recent one than the previous read.
10.3.2.2.4 Monotonic Writes
This property guarantees the serialization of the writes by one process. A write
operation on a data object or item must be completed before any successive writes by
the same process. Systems that do not guarantee this property are notoriously hard
to program [23].
10.3.3 e ventual C onsistenCy
In a replicated storage system, the consistency level defines the behavior of diver-
gence of replicas of logical objects in the presence of updates [24]. Eventual con-
sistency [23-25], is the weakest consistency level that guarantees convergence. In
the absence of updates, data in all replicas will gradually and eventually become
consistent.
10.3.3.1 System-Side Guarantees
Eventual consistency ensures the convergence of all replicas in systems that imple-
ment lazy, update-anywhere, or optimistic replication strategies [26]. For such sys-
tems, updates can be performed on any replica hosted on different nodes. The update