zip

[SOLVED] CSCE438 HW #5

35.00 $

Category:
Click Category Button to View Your Next Assignment | Homework

You will receive the following solution file(s) instantly after successful payment:

zip file icon HW5-4wyzo4.zip (126.5 KB)
Assignment Instructions Updated Recently? Submit Below and we will provide new Solution!
Submit New Instructions
🔒 Securely Powered by:
Secure Checkout
5/5 - (1 vote)
  1. [50 points (0.65pt each)] Choose the statements below that are TRUE:
    • Strict, Sequential and Causal Consistencies are Data-Centric consistency models.
    • Eventual Consistency, Monotonic Reads and Monotonic Writes are Client-Centricconsistency models.
    • Strict Consistency assumes absolute global time.
    • In Strict Consistency a “read is expected to return the value resulting from themost recent write”
    • Strict Consistency can be achieved if we have GPS.
    • In Sequential Consistency “the result of any execution is the same as if the (readand write) operations by all processes on the data were executed in an arbitrary order.”
    • In Causal Consistency, if an update U1 causes another update U2 to occur, then,U1 should be executed before U2 at each copy.
    • Causal Consistency is stronger than sequential consistency.
    • In Eventual Consistency, if no updates take place for a long time, all replicas willgradually become consistent.
    • In a Monotonic Read, “if a process reads the value of a data item x, then anysuccessive read operation on x by that process will always return that same value or a more recent value.”
    • In a Write Monotonic consistent store, a write operation by a process on a dataitem x is completed before any successive write operation on x by the same process.
    • One implementation of Sequential Consistency is to use a centralized process,called sequencer.
    • Write-through will impact caching in a distributed file system.
    • The read-ahead in distributed file systems, requests chunks of data when they areneeded.
    • The first version of NFS ran over UDP, using Sun RPC.
    • NFS uses caching at the client (caching data, file attributes and pathname bindings).
    • All NFS writes are write-through to disk.
    • Inconsistencies between local caches and server in NFS are solved by comparingtime-stamps.
    • NFS always invalidated data after some time (3 seconds for data in open files).
    • NFS version 2 extends the client-side buffer caching to disk, in 64KB chunks.
    • NFS version 3 continued to use UDP, because its simplicity.
    • NFS version 3 started to support 64bit file sizes.
    • GFS was designed with Google’s application workload specifically in mind.
    • In GFS, during a write operation, the master for a chunk sends data to replicasin a daisy chain.
    • In GFS, if the master reboots and then finds a chunck server has a newer versionnumber for a chunk, it adopts that version number.
    • In GFS, if a chunk server dies, then the master decrements the count of replicasfor all chunks on that chunk server.
    • GFS was designed for small streaming reads.
    • GFS was designed for large sequential writes that append.
    • In GFS, the performance of operations is very good for all apps.
    • GFS architecture contains one master server.
    • In the GFS architecture there is one chunk server for each chunk.
    • The master server in GFS holds metadata and most frequently accessed data files.
    • The master server in GFS holds all metadata in RAM.
    • A read/write operation with a chunk server in GFS specifies the chunk handleand the byte range.
    • In GFS, the client issues control/metadata requests to the chunk servers.
    • A client in GFS uses no caching.
    • The GFS consistency model defines “consistent” as: file region all clients see assame, regardless of replicas they read from.”
    • In GFS, the serial success result of a write operation is DEFINED.
    • In GFS, a successful record append data mutation is DEFINED.
    • In GFS, a successful concurrent write is CONSISTENT but DEFINED.
    • In GFS, a failure of a write or record append operation is INCONSISTENT.
    • BigTable is a sparse, centralized persistent multi-dimensional sorted map.
    • The map in BigTable is indexed by row key, column key and timestamp.
    • A row range (partition) in BigTable is also called tablet.
    • The items in a BigTable cell are stored in decreasing timestamp order.
    • BigTable is an alternate way for storing data than GFS and it implements its ownreplication.
    • BigTable processes share the same machines with MapReduce and GFS machines.
    • In the BigTable, there is a master server and many tablet servers.
    • In the BigTable, a client communicates directly with tablet servers for reads/writes.
    • The master server in BigTable maintains the set of live tablet servers and thecurrent assignment of tablets to tablet servers.
    • The memtable is an in RAM storage for storing the committed writes that arrivedat a tablet server.
    • When memtable size increases and reaches a threshold, it is frozen and committedto an SSTable in GFS.
    • The execution of a MapReduce program creates a Master process and severalWorker processes.
    • The master process of a MapReduce program assigns map and reduce tasks toworker processes.
    • The intermediate results of a MapReduce program are stored in GFS.
    • If the master process of a MapReduce program crashes a new master is electedthrough Chubby.
    • A Partition Function in a MapReduce program is used for ensuring that recordswith the same intermediate key end up at the same worker.
    • A MapReduce program will use information from GFS (location of replicas) todecide which file blocks will be processed by which worker process.
    • When a worker process in MapReduce segfaults, the information about the recordit processed is lost.
    • If the master process in MapReduce sees two failures for the same record, then ittells the next worker to skip the record.
    • In HDFS, a DataNode is the same as a ChunkServer in GFS.
    • In HDFS, a NameNode is the same as the Master Node in GFS.
    • A block in GFS is the same as a chunk in HDFS.
    • If a system can provide strong consistency, the programming model (how theprogrammer uses the API) is greatly simplified.
    • A Mobile Cloud consists of Distributed Storage and Distributed Data Processing
    • A k-out-of-n system is an n-component system that works if and only if k or lesscomponents work
    • Radio transmission is one of the major source of energy consumption on mobiledevices.
    • In “Resource allocation For edge computing” lecture data is allocated in a waythat the overall data retrieval energy is minimized.
    • In “Resource allocation For edge computing” lecture, the failure probability estimation includes failures due to disconnection from network.
    • In “Resource allocation For edge computing” lecture, Importance Sampling wasused for approximating the expected distance between two nodes.
    • The Read-ahead (prefetch) policy in a distributed file system, minimizes the waitwhen it actually is needed.
    • The Write-on-close policy in a distributed file system is synonym with sessionsemantics.
    • The VFS layer in NFS stands for Vectored File System.
    • In MapReduce, the arguments for a mapping function is a key and a value.
    • In MapReduce, the arguments for a mapping function is a key and a value.
    • The Partition Function in MapReduce ensures that records with the same intermediate key end up at the same worked.
  • HW5-4wyzo4.zip

Related products

Related in this category

More in this category