written 5.9 years ago by | modified 5.6 years ago by |
Mumbai University > Computer Science > Sem 8 > Parallel And Distributed System
written 5.9 years ago by | modified 5.6 years ago by |
Mumbai University > Computer Science > Sem 8 > Parallel And Distributed System
written 5.6 years ago by | • modified 5.6 years ago |
The consistency models are intended at delivering a system wide consistent view on a data store.
It is possible that concurrent processes might be concurrently updating the data store, and that it is essential to offer consistency in the face of such concurrency.
In object-based entry consistency, the data store warranties that when an object is called, the calling process is delivered with a copy of the object - that reflects all alterations or modifications to the object.
Throughout the call, it is also definite that no other process can inhibit that is to say, mutual exclusive access is delivered or offered to the calling process.
Sequential consistency might perhaps be guaranteed only when processes use synchronization mechanisms for instance locks or transactions; for performance reasons.
By presenting special client-centric consistency models, several inconsistencies might be hidden in a moderately inexpensive manner. That's why client centric consistency is needed.
Client centric consistency model:
1. Monotonic Reads
A data store is said to offer monotonic read consistency if the following condition holds true:
If a process reads the value of a data item x, any successive read operation on x by that process will always return that same value or a more recent value.
Monotonic read consistency assures that; if a process has seen a value of x at time t, it will never see an older version of x at any time later.
2. Monotonic writes
Most of the time, it is important that write operations occur in the precise and accurate sequence to all the replicas of the data store.
This scenario is assured in monotonic write consistency.
In a monotonic write consistent data store, the following conditions holds:
A write operation by a process on a data item x is completed before any successive write operation on x by the same process.
Basically it looks like data centric FIFO consistency. The essence of FIFO consistency is that write operations by the similar identical process are accomplished in the precise sequence universally.