midterm-sol - 15-440 Distributed SystemsMidtermName:Andrew:...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 15-440 Distributed SystemsMidtermName:Andrew: IDOctober 12, 2010Please write your name and Andrew ID above before starting this exam.This exam has 10 pages, including this title page. Please confirm that all pages are present.This exam has a total of 100 points.QuestionPointsScore1182731546525657681791Total:1001AShort Answers1. (18 points) True/False. Grading is +2 points for a correct answer and 0 points for either blank orincorrect. In other words, at the end of the exam, if you dont know the answer to any of these,guess,because youll get more points in expectation. If you want to be nice to the course staff, mark the onesyou guessed on with a G so that we have a better idea of what material to go over. We wont penalizeyou.True, False After being awakened fromcondwait(condition variable wait), a process must immediatelytry to acquire the lock associated with the condition variable.Solution:False: Upon awakening, the process is guaranteed to hold the lock.True, False In addition to greatly reducing the overall failure probability, moving from a single hard driveto using RAID 1 (mirroring) can also increase the throughput of small random reads.Solution:True: The small reads need only be sent to a single disk, so in theory, a RAID-1system can double small random read throughput.True, False RPC systems typically provide exactly-once semantics under arbitrary failures.Solution:False: RPC systemscannotprovide exactly-once semantics in all scenarios.They can either provide at-most-once or at-least-once.True, False The TCP transport protocol provides a reliable, in-order bytestream abstraction.Solution:True: Yes, yes it does.True, False In general, two computers can be time synchronized to within half of the round-trip delaybetween them.Solution:True: If you dont know more about the network between them, this is theaccuracy limit. (Imagine a completely asymmetric network with almost no delay fromA-B and almost all of its delay on the return trip from B-A.)True, False Kernel threads (typically) allow processes to harness multiple CPUs at the same time.Solution:True: They do indeed. This is one of the great things about kernel threads.True, False Switching between two kernel threads is faster than switching between two user threads becauseits done by the kernel.Solution:False: Its usually faster to switch between two user threads because you donthave the overhead of switching into the kernel and back.True, False A CPU must provide hardware support for atomic instructions in order for applications to beable to correctly synchronize.Solution:False: Its expensive, but we can use the distributed mutual exclusion tech-niques from lecture 11 to provide mutual exclusion without hardware support. But mostCPUs provide atomic instructions as a (very welcome) optimization and convenience forprogrammers....
View Full Document

Page1 / 10

midterm-sol - 15-440 Distributed SystemsMidtermName:Andrew:...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online