Modellering Tentaplugg

Question 1

A performance metric is used for what?

Answer 1

It is used to describe, measure and extrapolate information from data. Measure effectiveness and performance.

Question 2

What does a performance metric contain?

Answer 2

Count of how many times events occur, duration of time intervals, size of parameters. It also derives a value from these parts.

Question 3

A good performance metric has what characteristics?

Answer 3

Allows accurate and detailed comparisons, leads to correct conclusions, is well understood by everyone and has a quantitative basis. Also, a good metric helps avoid erroneous conclusions.

ADC | Accurate, Detailed, Correct

Question 4

What is a congestion collapse?

Answer 4

Offered load increases, whilst the work done decreases. A network becomes HIGHLY congested, leading to a significant drop in performance.

Question 5

Congestion collapses occur frequently in complex systems, but why does it happen?

Answer 5

Might happen because:
- The cost per job increases with the load
- Impatience, where jobs leave before completing
- Rejection, where jobs get rejected before completing

Question 6

How do we avoid congestion collapses?

Answer 6

With TCP congestion control and/or admission control in web servers

Question 7

What is a latent congestion collapse?

Answer 7

Think of it as a hidden congestion collapse, that has not yet collapsed. It is a fuse, ready to be set off. Some sort of bottleneck is preventing the collapse, but when resources gets added more and more, it might tip over and reveal the congestion collapse.

Question 8

What is response time?

Answer 8

The elapsed time from request, to response

Question 9

What is throughput?

Answer 9

Jobs / Operations completed per time unit

Question 10

What is bandwidth?

Answer 10

Bits per second

Question 11

What is Ad hoc metrics?

Answer 11

Metrics that will be defined for specific purposes

Question 12

What is jitter?

Answer 12

Variation in time delay

Question 13

What is delay?

Answer 13

Time taken to complete an action

Question 14

What is resolution?

Answer 14

Smallest increment between two possible measurements

Question 15

What is accuracy?

Answer 15

How close the mean of the distribution of measurements is to the true value

Question 16

What is precision?

Answer 16

The level of repeatability and consistency in measurements

Question 17

What does variability tell us? And what is variability also known as?

Answer 17

It tells us how far apart all the measured values are from eachother.

Variability is known as spread, or scatter

Question 18

How can we quantify variability?

Answer 18

Histograms: Equally sized buckets of observed data, with 6-7 buckets minimum
Box plots: A type of graph where each box has a line, showing max val, min val, quartiles and mean
Sample variance:
Standard deviation:

Question 19

What is variance?

Answer 19

A measurement that represents the deviation of observed values, from expected values. A high variance indicates greater dispersion between values, whilst a low variance suggests that the datapoints are closer to the mean.

Question 20

What is standard deviation?

Answer 20

A statistic that measures the dispersion of a dataset relative to its mean, and is calculated as the square root of the variance

Question 21

What is a QQplot?

Answer 21

A graphical tool used to compare the distribution of a dataset, to a theoretical distribution, or another dataset. It helps assess whether or not the data come from a population with a known or assumed distribution, such as a normal distribution

Question 22

Describe Full System Virtualization

Answer 22

FSV involves creating a complete and independent virtual instance of an entire computer system (this includes operating system, hardware components, software, etc.). This environment mimics the original system closely.

Question 23

What does FSV allow the user to do? One point, important

Answer 23

It allows the user to run multiple operating systems on a single machine.

Question 24

Describe Lightweight System Virtualization

Answer 24

LSV involves creating isolated environments (called containers) that share the same operating system kernel as the host system. These containers are lightweight, portable and efficient in resource utilization.

Question 25

What are the LSV isolated environments known as?

Answer 25

Containers.

Question 26

Does mininet use LSV or FSV?

Answer 26

LSV

Question 27

Simulation is the process of what?

Answer 27

1. Designing a model of a real system
2. Conducting experiments with this model

Question 28

Why do we simulate?

Answer 28

To gain an understanding of the behavior of the system.

Question 29

What is a continuous-time model?

Answer 29

A model where state changes occur continuously across time

Question 30

What is a discrete-time model?

Answer 30

A model where state changes occur only at specific times

Question 31

Describe the Monte Carlo Simulation

Answer 31

It involves generating numerous random inputs within specified ranges and using them to simulate the system’s outcomes. When this has ran several times, it provides a probalistic view of possible outcomes and helps in making informed decisions.

Note that the name comes from Monte Carlo Casino, making it easy to remember its’ relevance in chance and randomness.

Question 32

What’s the best and worst with conducting Live Experiments?

Answer 32

They provide high accuracy but are expensive, and less flexible.

Question 33

In what does emulation strike a balance?

Answer 33

It strikes a balance in cost, accuracy and flexibility.

Question 34

What’s the best and worst with conducting simulations?

Answer 34

They are really cost effective, and highly flexible, but may have accuracy limitations based on the model

Question 35

What is analytical modelling?

Answer 35

A strictly theoretical analysis based on a mathematical and/or statistical approach.

Question 36

What are all the 4 types of means?

Answer 36

Arithmetic mean, harmonic mean, geometric mean and weighted mean

Question 37

When should you use arithmetic mean?

Answer 37

Arithmetic mean is used when you want to find a typical or average value of a set of numbers, where all values have equal importance.
Ex: Used when calculating average exam scores in class

Question 38

When should you use harmonic mean?

Answer 38

Used when you want to calculate an average that considers rates, ratios, or speeds, especially when dealing with reciprocals or inversely proportional quantities. Commonly used in physics, economics and engineering for averaging rates or speeds.
Ex: Calculating average speed in a round-trip journey.

Question 39

When should you use geometric mean?

Answer 39

Used when you want to calculate a central value for a set of numbers that are subject to exponential growth or decay. Could be used in finance to calculate average growth rates, such as investment returns over multiple periods.
Ex: Calculating average annual growth rates of investments.

Question 40

When should you use weighted mean?

Answer 40

Used when you want to calculate a central value for a set of numbers, but each value has a different level of importance or weight. It’s used in various scenarios where certain values carry more significance.
Ex: Calculating the GPA of a student with different credit weights for courses.

Question 41

Är du självmordsbenägen?

Answer 41

Ja

Question 42

Why do queues exist?

Answer 42

Since jobs share many resources and only one can access at a time, then the other must wait. While waiting, they queue up.

Question 43

What is the Kendall Notation, and what does it describe?

Answer 43

The Kendall Notation is the standard system used to describe and classify a queueing node.

Question 44

What can be said about the Kendall Notation: M/M/3/20/1500/FCFS?

Answer 44

• Exponentially distributed arrivals
• Exponentially distributed service times
• Three servers
• Capacity 20 (queue size is 20 - 3 = 17)
• Population is 1500 total
  -Service discipline is FCFS

Question 45

What is the formula of Little’s Law?

Answer 45

Mean jobs in system = arrival rate * mean response time

Question 46

In the Kendall Notation, what does “M” and “D” stand for?

Answer 46

M = Exponential / Memoryless / Poisson distribution
D = Deterministic distribution

Question 47

What is the Simpson’s Paradox?

Answer 47

Simpson’s Paradox, also known as confounding, is a phenomenon in statistics where a trend appears in different groups of data, but it reverses or disappears when the groups are combined.
In laymans terms, the overall pattern can be misleading or contradictory when looking at the data as a whole versus when it’s broken down into subgroups.
This can happen when the subgroups have different sizes or weights that influence the combined results.

Question 48

What is verification and validation in systemmodelling?

Answer 48

To validate a model is to analyse if it is a correct representation of the real system.

To verify a model is to see if the model behaves as expected.

Question 49

How can validation be performed?

Answer 49

Validation can be performed by comparing the model to other performance evaluation approaches. For example, a few measurements could be done on the real system and compared to the model to check its validity.

Question 50

How can verification be performed?

Answer 50

Verification can be performed by doing some different tests on the simulation. Slightly changed inputs should yield a slight change of output.

Question 51

What is ANOVA?

Answer 51

It’s a statistical technique used to analyze the differences among various groups or treatments in an experiment. In simple terms, ANOVA helps you determine whether there are significant differences between the groups you’re studying and if those differences are due to the factors you’re testing.

It assesses whether the differences among groups are greater than what you would expect from random chance. In other words, it tells you if the variations you see are likely due to the factors you’re investigating and not just random fluctuations.

Question 52

How is ANOVA performed?

Answer 52

First do an equal amount of repeated measurements on a number of different systems. Store the measurements in a table

The table has columns representing a system each, and rows representing a repetition each.

1.Calculate group means: Find the average (mean) of the data for each group. This tells you the typical performance for each system.

2. Calculate overall mean: Find the average of all the data points combined, regardless of their group. This is the overall average performance.
3. Calculate variability: Determine how much each data point varies from the group mean, and then square these differences.
4. Perform the ANOVA test: This involves comparing the variability within each group (due to random chance) to the variability between groups (due to the different treatments). You can use statistical software or tables to do this.
5. Determine statistical significance: ANOVA will provide a p-value. If this value is less than a predetermined significance level (usually 0.05), it suggests that there are significant differences between the groups.

Question 53

How are ANOVA contrasts useful?

Answer 53

The use of contrast gives the ability to compare a subset of the systems to see if there is a significant difference within the subset.

Question 54

What is a Discrete-Event Simulation?

Answer 54

A DES is time based, where the incrementation of time will cause events to execute. An example of this is a simulation of processes arriving at a server. The processes are arriving with a specified rate, which is dependent on the time.