Consider the Big-Data processing systems Spark, Spark Streaming and Flink.

Taking into account their semantics, fault-tolerance handling and input management, in order to achieve low latency even in the presence of faults, identify the true comparative assessments about them from the following.

(select all the correct options).

In the context of tabular cloud storage, consider DynamoDB whose design is mainly directed towards scalability and good performance.

Because of its design and goals, only eventual consistency is offered by default, which may incur in update conflicts.

What is the mechanism provided by DynamoDB that can help programmers avoid/reduce such conflicts?

(select all the correct options).

CNV-25-26 Exam 2:

Identification and Rules:

Please fill in your identification in the draft sheet you were provided. Return it at the end of the exam after completely closing your submission.

In the draft sheet and in this question you must insert your exam code given to you by the person watching the exam. You exam is not valid without the exam code.

The exam is without consultation (no other windows or apps beside the one where the exam is being done). Otherwise, exam is annulled.

The exam has a time limit of 80 minutes. You can only leave after submitting your exam and only after 45 minutes. We take the best grade of the two exams.

There will be no further information about the exam questions (dúvidas acerca do enunciado) provided during the exam, besides the question text itself.

Answer the questions given the context in the question.

There are 24 actual questions in total in the exam, all graded with the same weight except the Map-Reduce problem that has a weight twice a normal (all others) question (1 mark -> 0.8 in 1-20 scale). 

E.g. for a final grade of 8/20 ("valores") you need to sum up 10 marks, for a final grade of 15/20 ("valores") you need to sum up 18.75 marks. For a grade of 20/20 ("valores"), you need to sum up the 25 complete marks.

(Only if strictly needed, you can also insert here any other notes regarding the answering of some question(s) in the exam clearly identifying them in your comments). 

In the context of System VMs, performance can be enhanced resorting to techniques such as paravirtualization and hardware extensions (such as VT-x Vanderpool) to support virtualization. 

Which sentence is true when considering improvements to emulation performance in System VMs?

(select the option that best fits the question).

In AWS EC2, each instance has an associated state in each moment regarding its current operation. 

What are the examples containing possible EC2 instance states?

(select all the correct options).

Operand Stack Tracking is performed at class loading to check the validity of the byte-code, by tracking all possible control flows and assessing whether the operand stack always keeps a valid configuration. 

Consider the following excerpt representing the byte-code of a method:

1:  iload_1

2:  iload_2

3:  if_icmplt 6

4: iconst_0

5: goto 7

6: iconst_1

7: iadd

8: ireturn

Determine and briefly explain whether this byte-code would pass or fail the check during operand stack tracking. 

If you consider it would pass, state PASS, and briefly explain why.

If you consider it would fail, state FAIL, identify the byte-code line where that check would fail and briefly explain why.

(No penalty for wrong answer.)

In the context of Cloud Computing, computing resources are regarded as an utility. What are the advantages that cloud computing offers?

(select all the correct options).

Flink aims at providing stream processing with low latency, high throughput and fault-tolerance.

When considering Flink and other Big-Data processing systems, regarding fault-tolerance, identify the true statements.

(select all the correct options).

Function-as-a-Service (FaaS) aims to improve on the delivery model provided by Infrastructure-as-a-Service (IaaS). 

Identify the key advantages that FaaS can bring to customers deploying applications on the cloud, when compared to IaaS.

(select all the correct options).

Consider a cloud deployed application storing and processing large volumes of data (Big-Data), e.g. in order to process historical data such as logs of credit card and bank transfer operations.

In this scenario, what properties can the application expect from typical cloud storage systems such as Google File System/HDFS and BigTable?

(select the option that best fits the question).