Unit 1: Virtualization and Distributed Computing - Practice Quiz

1 What is virtualization in computing?

2 Which software layer creates and manages virtual machines?

3 What is a virtual machine?

4 What does x86 virtualization primarily allow?

5 Which processor feature supports hardware-assisted x86 virtualization?

6 Which type of virtualization creates virtual versions of complete computer systems?

7 Which type of virtualization provides users with remotely hosted desktop environments?

8 What does storage virtualization combine?

9 What does VLAN stand for?

10 What is a key benefit of using VLANs?

11 What does VSAN commonly stand for in storage virtualization?

12 Which is a common benefit of a VSAN?

13 What is distributed computing?

14 How do computers in a distributed system usually communicate?

15 What is a feature shared by parallel and distributed systems?

16 What is the main goal of parallel computing?

17 Which component is typically present in a multicore parallel computer?

18 In a shared-memory architecture, what do processors commonly access?

19 What is a node in a distributed system?

20 Which statement correctly compares parallel and distributed computing?

21 A company wants to consolidate several underutilized servers onto one physical machine while allowing each server workload to run its own operating system. Which virtualization component is primarily responsible for this?

22 A host has 32 GB of physical memory, but its virtual machines are configured with a total of 48 GB. Under which condition can this configuration operate effectively?

23 An early x86 processor executes some sensitive guest instructions without trapping to the hypervisor. Which technique can a hypervisor use to virtualize such a guest safely?

24 A virtualization administrator enables Extended Page Tables (EPT) on an Intel processor. Which operation is most directly accelerated?

25 A guest operating system is modified so that it replaces privileged operations with explicit calls to the hypervisor. Which type of virtualization is being used?

26 A development team needs hundreds of isolated application environments that start quickly and use minimal memory. All applications can use the same operating system kernel. Which approach is most suitable?

27 Two departments use the same physical Ethernet switches but are assigned to different VLANs. By default, what is required for a host in one VLAN to communicate with a host in the other?

28 A storage administrator divides one physical Fibre Channel fabric into multiple VSANs. What is the main operational benefit?

29 A distributed application accesses a remote object using the same style of method call used for a local object. Which distributed-computing goal does this illustrate?

30 A service consists of five networked nodes. One node becomes unreachable, while the remaining nodes continue processing requests. Which characteristic of distributed computing does this scenario demonstrate?

31 Which workload is best classified as parallel rather than primarily distributed?

32 A computation is divided among networked computers that do not share memory. Which mechanism is normally required for the computers to exchange intermediate results?

33 A program spends 80% of its execution time in code that can be perfectly parallelized. According to Amdahl's law, what is its theoretical speedup when that portion runs on four processors?

34 A large image is divided into equal-sized regions, and the same blur operation is applied to each region simultaneously. Which parallelization strategy is being used?

35 On a NUMA server, a thread repeatedly accesses a large memory region. Which placement generally provides the best performance?

36 Two processor cores update different variables that happen to occupy the same cache line. Performance drops because the line repeatedly moves between caches. What is this problem called?

37 A replicated database has five replicas and uses a write quorum of three and a read quorum of three. Why can a read obtain at least one replica that participated in the latest successful write?

38 In a system using Lamport logical clocks, event causally precedes event . What must be true about their timestamps?

39 Several universities contribute heterogeneous computing resources to execute large scientific jobs under different administrative domains. Which computing paradigm best matches this arrangement?

40 A retailer experiences unpredictable traffic spikes and wants servers to be provisioned automatically and billed according to usage. Which paradigm is the best fit?

41 An early x86 guest executes POPF after being deprivileged to a nonzero privilege ring. The instruction does not trap but silently preserves privileged flag bits, preventing conventional trap-and-emulate virtualization. Which technique most directly preserves guest semantics?

42 A hypervisor changes an Extended Page Table entry from writable to read-only while a vCPU is paused. Why may updating the EPT entry alone be insufficient before resuming that vCPU?

43 A latency-sensitive VM receives a high rate of device interrupts. Which hardware-assisted design most directly reduces VM exits when the target vCPU is already running?

44 A disk-only VM snapshot is taken while a database transaction is updating several blocks in memory and on disk. Which statement best characterizes restoration from that snapshot?

45 A Linux host must simultaneously run an unmodified Linux guest and an unmodified Windows guest, each with its own kernel. Which virtualization approach satisfies this requirement?

46 Compared with a fully emulated network adapter, a paravirtualized network driver primarily improves performance through which mechanism?

47 Two IEEE 802.1Q trunks use native VLAN 10 on one endpoint and native VLAN 20 on the other. What is the most significant consequence for untagged frames crossing the trunk?

48 Hosts in VLAN 40 on two different switches cannot exchange even broadcast traffic. Access ports are correctly assigned, but VLAN 40 is absent from the allowed list on the inter-switch trunk. Which change is sufficient without altering broadcast-domain boundaries?

49 A vSAN object uses a RAID-1 policy with failures to tolerate set to 1. Assuming the standard witness-based layout, which component arrangement is required?

50 A vSAN administrator selects RAID-6 erasure coding with failures to tolerate set to 2. What is the standard minimum component structure for the protected data stripe?

51 In a distributed execution, Lamport timestamps satisfy . Which conclusion is valid without additional information?

52 A proposed global snapshot includes the receipt of message by process but excludes the sending of by process . Why is this snapshot not a consistent cut?

53 A bulk-synchronous parallel computation has balanced local workloads, but one worker intermittently pauses for garbage collection before each barrier. What is the dominant system-level effect?

54 A program has a strictly serial fraction of , while the remaining fraction parallelizes perfectly. Ignoring overhead, what is its maximum speedup on 32 processors according to Amdahl's law?

55 A parallel floating-point reduction produces slightly different sums when the thread count changes, even though no data race exists. What is the most likely cause?

56 A computation DAG has total work and span . Ignoring scheduling overhead, which bound applies when it runs on 16 processors?

57 Two cores repeatedly update different counters that occupy the same cache line. Profiling shows heavy coherence traffic despite no logical data sharing. Which architectural effect explains this behavior?

58 Under the MESI protocol, Core A holds a cache line in the Modified state. Core B then issues a read miss for that line. What state transition normally follows?

59 A replicated store has replicas. It requires quorum read-write overlap, quorum write-write overlap, and write availability despite any two unavailable replicas. Which choice maximizes read availability under these constraints?

60 A fixed cluster runs an iterative fluid simulation in which neighboring subdomains exchange boundary values after every short compute phase. Which paradigm is generally the best fit?