0% found this document useful (0 votes)
20 views

ESLE - Sample Exam

Design B is expected to perform better than Design A based on the given scalability curves. Design B scales better since its throughput increases more sharply with the number of nodes. Batching allows for further optimizations by reducing the number of requests sent and amortizing overhead costs across multiple requests. The rule of thumb for capacity planning predictions based on the Universal Scalability Law is to limit estimations to double the actual throughput or size to account for uncertainty in the model as system scale increases.

Uploaded by

Gonçalo Duarte
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
20 views

ESLE - Sample Exam

Design B is expected to perform better than Design A based on the given scalability curves. Design B scales better since its throughput increases more sharply with the number of nodes. Batching allows for further optimizations by reducing the number of requests sent and amortizing overhead costs across multiple requests. The rule of thumb for capacity planning predictions based on the Universal Scalability Law is to limit estimations to double the actual throughput or size to account for uncertainty in the model as system scale increases.

Uploaded by

Gonçalo Duarte
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Instituto Superior Técnico

Large-Scale Systems Engineering – Engenharia de Sistemas de Larga Escala – MEIC/METI

Sample Exam – DATE – Duration of the exam: 2 hours

Your answers must only use the number of lines in the boxes provided next to each question. If
necessary, for instance to correct a previous answer, you can use the space at the end of the exam
sheet but you cannot use more lines than in the original box. Justify all answers. Answers can be in
English or Portuguese.

Number Name

System Scalability

1. You are responsible to compare two alternative designs for a novel distributed system. After a preliminary
analysis of the competing designs, you identify the following scalability parameters. For design A, the
parameters are λ = 80; σ = 0.05; 𝑘 = 0.01 and for design B the parameters are λ = 160; σ = 0.05; 𝑘 = 0.01.
The figure below presents the scalability curves for both systems. Which design is expected to perform
better? Which design scales better?
700
Design A (l=80, d=0.05, k=0.01)
Design B (l=160, d=0.05, k=0.01)

600

500
Throughput (ops/s)

400

300

200

100

0
0 2 4 6 8 10
N
2. A marketer is trying to sell ImageX - an image hosting system - to be deployed at your company.
The workload is composed of millions of static images of the same size. He claims that ImageX provides
constant latency, regardless of the number of nodes required to host all images. Do you advise the CEO of
the company to acquire ImageX?

System Performance
3. Discuss two techniques studied in the classes to reduce latency in a request processing pipeline.

4. Batching is a technique that groups several requests into one with the goal of amortizing the cost of
sending single requests between different stages. Which further optimizations are possible when using
batching?
Number Name

Benchmarking
5. Consider the experimental design below with 7 factors. Which conclusions can you draw?

A.B A.C B.C A.B.C


Exp. A B C D E F G
1 -1 -1 -1 1 1 1 -1
2 1 -1 -1 -1 -1 1 1
3 -1 1 -1 -1 1 -1 1
4 1 1 -1 1 -1 -1 -1
5 -1 -1 1 1 -1 -1 1
6 1 -1 1 -1 1 -1 -1
7 -1 1 1 -1 -1 1 -1
8 1 1 1 1 1 1 1
%SST 25 12 15 5 6 32 5

6. What is the unidirectionality of an effect? How important is it when creating an experimental design?
Capacity Planning
7. A rule of thumb when doing predictions based on the Universal Scalability Law is to limit the estimation to
the double of the actual throughput or the double of the actual system size. What is the underlying reason for
such rule?

8. What is the role of Effective Demand in Capacity Planning?

Self-Adaptation
9. Describe two self-adapting properties and provide concrete application examples where they can be used.
Number Name

10. In the MAPE-K framework, what is the role of the Plan stage?

You might also like