55 Questions
What is the primary focus of the Distributed File System (DFS) lesson?
Caching mechanisms
In a Distributed File System, which nodes own and manage the state?
Server nodes
What is the characteristic of a Peer Distributed Application?
All nodes are peers and share the state
What is the overall application state in a Peer Distributed Application?
The union of all local states
What is the relationship between nodes in a Peer Distributed Application?
Peer-to-peer relationship
What is a characteristic of state in a Peer Distributed Application?
Some state is locally stored on a particular node
What is the issue with using variables in a distributed shared memory (DSM) system?
Variables are still too small and occur too frequently
What is a more suitable granularity for a DSM system integrated at the operating system level?
Page-based
What is the advantage of using page-based granularity in a DSM system?
It amortizes the cost of remote access across larger granularities
What is a disadvantage of the language level solution for DSM?
It is not generalizable outside of programming languages with DSM support
What is the issue with increasing the granularity of sharing in a DSM system?
It may lead to false sharing
What is an example of false sharing in a DSM system?
Two processes accessing different variables on the same page
What is the main role of the DSM layer in a single reader/single writer application?
To provide the application with the ability to access additional, remote memory
What is the performance metric to analyze in a DSM solution?
Access latency
What is a way to maximize the proportion of local memory accesses in a DSM system?
Through migration
What is the limitation of the migration solution in a DSM system?
It incurs overheads
What is the primary reason for creating page replicas in datacenter environments?
To triplicate shared state
What is the role of the manager node in a DSM system?
To maintain metadata for a particular page
What is the purpose of the global map in a DSM system?
To capture the page ID and return a manager ID
Why does the DSM layer need to intercept every access to the shared state?
All of the above
What is the role of the MMU in a DSM system?
To generate a trap when a valid mapping is not found
What is the benefit of using a global mapping table in a DSM system?
It allows for manager node flexibility
Why is it important to maintain metadata for each page in a DSM system?
To perform specific access to that page
What is the advantage of using an object-based DSM system?
It can be implemented at the level of the programming language runtime
What is the purpose of the DSM layer in a DSM system?
To intercept every access to the shared state
What guarantee do causal consistency models provide in terms of update ordering?
Correct ordering for causally related updates
In the weak consistency models, what purpose does a synchronization point serve?
Ensuring visibility of previous updates at other processors
What does a synchronization operation performed by P1 after writing to m1 guarantee according to the weak consistency model?
Visibility of previous updates
How does the weak consistency model handle synchronization between processors regarding visibility of updates?
By requiring explicit synchronization calls from relevant processors
What is the primary purpose of a sync operation in weak consistency models?
Ensuring the visibility of past updates to all processors
What is the consequence of not explicitly synchronizing with the rest of the system in weak consistency models?
Updates may not be immediately visible
In weak consistency models, when does P2 see all previous updates made to any memory location in the system?
After a synchronization operation on P2
What controls the overheads imposed by the DSM layer in fine-grained synchronization operations?
Different synchronization operations for different granularities
What is the purpose of an entry/acquire point in a system providing finer-grained synchronization operations?
To make all updates visible to a particular process
How does separating synchronization types into entry/acquire and exit/release points benefit the system?
Limits data movement and coherence operations
Why does the solution of permitting multiple copies of the same data page require consistency management?
To ensure data is propagated across nodes for up-to-date reads
What is the purpose of consistency management in DSM systems?
To maintain consistency similar to SMP systems
What is the strategy called when one processor’s cached value of a variable is invalidated when another processor updates that variable?
Write-invalidate
What is the drawback of supporting the write-invalidate strategy in a DSM system?
The overhead is too high once network costs are considered
What is the purpose of the home node in a DSM system?
To drive all of the coherence operations related to a page
What is the role of the owner node in a DSM system?
To currently own the page and control all of the state modifications
What is the purpose of separating the distinction of home node from an owner node in a DSM system?
To allow for efficient handling of repeated access to a page by a node that is not its home node
What is the purpose of caching in a DSM system?
To achieve low latency and performant system
What is the global shared memory composed of in a DSM system?
The memory pages that each node contributes towards the DSM system
What is the purpose of the DSM layer in a DSM system?
To incorporate caching to achieve low latency
What is the primary purpose of a consistency model in a distributed system?
To provide a guarantee that state changes will behave in a certain way
What is the main constraint of the interleaving in sequential consistency?
Updates from the same process will not be arbitrarily interleaved
What is the main difference between strict consistency and sequential consistency?
Strict consistency guarantees immediate visibility, while sequential consistency does not
Why is strict consistency impossible to achieve in practice?
It is impossible to guarantee ordering of accesses from different cores without locking and synchronization
What is the main advantage of sequential consistency over strict consistency?
It is more suitable for distributed systems
What is the main goal of causal consistency?
To allow processes to see different updates based on their dependencies
What is the main limitation of strict consistency?
It is impossible to guarantee instantaneous updates
What is the relationship between the consistency model and the software layers in a system?
The software layers must follow certain rules to achieve the consistency model
Why is it important to ensure that the way operations access memory is representative of how the operations were issued?
To ensure that memory behaves correctly
What is the main advantage of consistency models in distributed systems?
They provide a guarantee that state changes will behave in a certain way
Test your knowledge on distributed file systems and their caching mechanisms. Explore how files are stored on server nodes and accessed by client nodes in a distributed environment.
Make Your Own Quizzes and Flashcards
Convert your notes into interactive study material.
Get started for free