Quiz 8 Flashcards
(12 cards)
Some reasons for replication can be:
Increase reliability
Improve availability
Improve usability
improve performance
Increase reliability
Improve availability
improve performance
The factors to be considered when we relax consistency are:
Read and write patterns of the replicated data
The number of replica servers available for replicated data item
The purpose for which the data is used
The type of communication protocols available for performing updates
Read and write patterns of the replicated data
The purpose for which the data is used
Continuous consistency model can provide:
eventual consistency of replicated data
what ordering of operations to be ensured at the replicas
a way to measure and specify the level of inconsistencies that an application can tolerate
a way to measure and specify the level of inconsistencies that an application can tolerate
Monotonic-read consistency model guarantees that:
Updates are propagated as a result of previous write operations
A write operation by a process on a data item x is completed before any successive write operation on x by the same process
Once a process has seen a value of x, it will never see an older version of x
Once a process has seen a value of x, it will never see an older version of x
A client-centric consistency model where a client process can modify a password on a server replica and the password change is seen immediately by the same client on a different replica server can be said to implement:
Monotonic-read model
Monotonic-write model
Read your writes model
Writes follow reads model
Read your writes model
In remote-write protocols:
All write operations need to be forwarded to a single server
Operations are ordered in a sequential order
Every data item has a primary server assigned to it
All write operations need to be forwarded to a single server
Operations are ordered in a sequential order
Every data item has a primary server assigned to it
The main challenge with active replication is that:
Events have to be ordered across all replicas
Communication may not be reliable
It is not fault tolerant
Events have to be ordered across all replicas
A replicated-write protocol which is based on majority voting is called:
Coherence-based protocol
Active replication
Local-write protocol
Quorum-based protocol
Quorum-based protocol
One way to enforce cache coherence when shared data is cached locally is to let the server send an invalidation to all caches when a data item is modified
True
False
True
In replicated-write protocols:
Write operations can be performed at multiple replicas
All of the above
Write operations can be performed in only one replica
Write operations can be performed at multiple replicas
Client-centric consistency protocol is about
Providing consistency for a single client that accesses different replicas
All of the above
Making sure that there is a systemwide consistent view on a data store
Providing consistency for a single client that accesses different replicas
In primary-based protocols:
Updates can be performed at multiple replicas
Updates can be performed in only one replica
All of the above
Updates can be performed in only one replica
