Unit 3: Cloud Architecture and Economics - Practice Quiz

1 Which layer of the cloud computing stack provides complete applications to end users over the internet?

2 Which cloud service model commonly provides virtual machines, storage, and networking resources?

3 What is the main purpose of workload distribution in a cloud environment?

4 Which component commonly distributes incoming requests among multiple servers?

5 What is the primary goal of capacity planning?

6 What is a likely result of provisioning too little capacity for an application?

7 What does cloud bursting allow an organization to do?

8 Which situation is most suitable for cloud bursting?

9 What is thin provisioning?

10 How does thick provisioning allocate storage?

11 In Service-Oriented Architecture, an application is primarily organized as a collection of what?

12 What allows users or other services to interact with an SOA service?

13 What is a Service Level Agreement (SLA)?

14 In an SLA, what does a target of usually describe?

15 Which statement best reflects the cloudonomics principle that on-demand resources can outperform forecasting?

16 According to cloudonomics, why can sharing cloud infrastructure among many customers improve efficiency?

17 A cloud service charges per hour. What is the estimated cost of using it for hours?

18 Which basic formula can be used to estimate the cost of a usage-based cloud resource?

19 Which pricing option is often economical for a predictable workload that will run for a long period?

20 Which economic strategy helps avoid paying for cloud resources that are no longer needed?

21 A company wants developers to deploy applications without managing operating systems, while still controlling application code and data. Which cloud service layer best fits this requirement?

22 An organization moves from IaaS virtual machines to a SaaS accounting product. Which responsibility most clearly shifts from the organization to the provider?

23 A web application has six identical instances, but two receive most requests because clients reuse long-lived connections. Which change best improves workload distribution?

24 Users in Asia and Europe access the same stateless service. The goal is to reduce latency while retaining regional fault tolerance. Which approach is most suitable?

25 A service handles 1,200 requests per second at 60% CPU on eight identical instances. Assuming linear scaling, how many instances are needed to handle 1,800 requests per second at approximately the same utilization?

26 A retailer sizes its cloud environment using average daily demand, but checkout latency becomes unacceptable during short sales peaks. Which planning change best addresses the problem without permanently provisioning for the maximum?

27 A private cluster normally runs at 70% utilization but exceeds capacity for two hours at month-end. Which workload is the best candidate for cloud bursting?

28 Before bursting a data-intensive analytics job from a private cloud to a public cloud, which factor is most important to evaluate?

29 A team creates a 4 TB virtual disk for a new application but expects only 500 GB of initial usage. Which provisioning method minimizes immediate physical storage consumption?

30 Several thin-provisioned volumes have logical capacities totaling 50 TB on a 20 TB storage pool. What is the main operational risk?

31 An order-processing service must work with multiple payment providers without being rewritten for each provider. Which SOA principle most directly supports this goal?

32 A composite travel-booking process calls flight, hotel, and payment services. The payment service fails after the hotel is reserved. Which SOA mechanism best handles this situation?

33 An SLA promises 99.9% monthly availability. For a 30-day month, approximately how much downtime is allowed?

34 A provider reports 99.95% availability, but the SLA excludes planned maintenance. What should a customer verify before comparing this figure with another provider?

35 A news site experiences rare, unpredictable traffic spikes that are ten times its normal load. Which cloudonomics idea most strongly favors renting elastic cloud capacity?

36 Two independent services each have variable demand peaks at different times. Why can hosting both in a shared cloud resource pool reduce required capacity?

37 A workload uses 10 instances for 120 hours. Each instance costs $0.20 per hour, and storage costs $30 for the period. What is the estimated total cost?

38 A migration estimate includes compute and storage charges but predicts a much lower bill than the first production month. The application sends large reports to users outside the cloud. Which omitted item most likely explains the difference?

39 A company has a stable baseline workload for three years and unpredictable seasonal peaks. Which purchasing strategy is generally most economical?

40 A startup is comparing an on-premises system with a cloud service. The cloud's direct monthly cost is slightly higher, but demand is uncertain and the product may be discontinued within six months. Which factor most strengthens the cloud option?

41 An application is migrated from a PaaS to raw virtual machines in an IaaS environment. The application previously relied on the PaaS for runtime patching, autoscaling, and a managed message queue. Which responsibility shift must the architecture explicitly address?

42 A service requires a custom kernel module for packet processing but uses a standard relational database. Which deployment best minimizes operational responsibility while preserving the kernel requirement?

43 A load balancer distributes 800 requests per second across two pools. Pool X has four instances rated at 100 requests per second each, while Pool Y has two instances rated at 300 requests per second each. Which routing split equalizes utilization across all instances?

44 A consistent-hashing cache cluster contains nodes with capacities in the ratio . Equal numbers of virtual nodes currently cause the largest node to remain underused. Which change best preserves low remapping while correcting the imbalance?

45 A shared service has a peak demand of 150 compute units. Each instance provides 20 units, but policy limits planned utilization to 75%. The service must still meet peak demand after any one instance fails. What is the minimum number of instances?

46 Two workloads each have mean demand 100 units and standard deviation 20 units. Their correlation coefficient is 0.75. Assuming joint normality, approximately what aggregate capacity covers demand at the one-sided 95th percentile, using ?

47 A burst requires additional resource-hours. Each unserved resource-hour causes loss , public-cloud usage costs per resource-hour, and activating the burst incurs fixed cost . Ignoring future effects, when is bursting economically justified?

48 A private-cloud application must burst during demand spikes, but regulation prohibits customer records from leaving the private region. Which architecture most directly satisfies the constraint while still adding public-cloud capacity?

49 A thin-provisioned pool has 30 TB of physical capacity and an operational threshold of 80%. One hundred VMs initially consume 0.20 TB each. VM data grows by 0.010 TB per VM per day, and retained snapshots add 0.002 TB per VM per day. After approximately how many days is the threshold reached?

50 A workload requires 6,000 random 8 KiB IOPS. One provisioned disk supports at most 5,000 IOPS and 125 MiB/s. Assuming ideal RAID 0 scaling and no redundancy requirement, what is the minimum disk count imposed by these limits?

51 An order workflow spans inventory, payment, and shipping services. A global ACID transaction is unavailable, and shipping cannot always be reversed after dispatch. Which design best reduces inconsistent outcomes?

52 A provider adds an optional field to an SOA response. Some consumers fail because they reject every unknown field. Which contract strategy most effectively supports independent service evolution?

53 A request succeeds only if two independent services are available. Their availability SLAs are 99.95% and 99.90%. What is the resulting end-to-end availability, and does it satisfy a 99.90% end-to-end target?

54 A 30-day month contains 43,200 minutes. An SLA excludes scheduled maintenance from both downtime and the measurement denominator. The service has 25 minutes of total unavailability, including 6 minutes of scheduled maintenance. Does it meet a 99.95% availability SLA?

55 Workload A requires 100 units in period 1 and 20 in period 2, while Workload B requires 20 units in period 1 and 100 in period 2. Compared with separately provisioning each workload for its peak, how much capacity does perfect pooling save?

56 One unit of reserved capacity costs $0.07 per hour for every committed hour, whether used or idle. Equivalent on-demand capacity costs $0.12 per used hour. Ignoring risk and flexibility value, above what utilization does the reservation become cheaper?

57 A three-year private deployment costs immediately and at each year-end, with a residual value at the end of year 3. A cloud alternative costs at each year-end. At a 10% discount rate, which alternative has the lower present cost?

58 A cloud workload stores an average of 50,000 GB at $0.02 per GB-month, makes 20 million requests at $0.004 per 10,000 requests, and transfers 8,000 GB out. The first 1,000 GB of egress is free and the remainder costs $0.09 per GB. What is the estimated monthly cost?

59 A fault-tolerant batch workload can run on discounted interruptible instances from two markets. The markets have slightly different prices and partially correlated interruption events. Which strategy best minimizes expected completion cost when missed deadlines are expensive?

60 A company expects demand to become clear in one year. Building private capacity now is cheaper per unit but requires an irreversible investment sized for uncertain peak demand. What is the strongest economic reason to use elastic cloud capacity during the first year?