Föreläsning 4 - Quality Characteristics of IoT-Based Systems

Question 1

What does functionality mean in software architecture? -

Answer 1

• Functionality means what the system is supposed to do — its main tasks or purpose.
• It doesn’t control or determine the architecture. → You can build many different architectures to support the same functionality.
• Functionality and quality attributes (like performance, security, etc.) are separate ideas — they don’t affect each other directly.

Question 2

What are quality attribute considerations? -

Answer 2

If a functional requirement is “when the user is home, → thermostat should heat the apartment if temp is below…”. Quality attribute:

• Performance – How fast it heats.
• Availability – How often it fails & how fast it’s fixed.
• Usability – How easy it is to learn user preferences.

Question 3

Three problems with quality attributes and solutions -

Answer 3

• Definitions aren’t testable (saying a system is “modifiable” means little without specifics).
• Overlapping concerns (It’s unclear which attribute an issue belongs to (e.g., system failure from an attack)).
• Different vocabularies (each quality attribute area uses its own terms).

Solutions: Use quality attribute scenarios to make attributes clear and testable. Discuss each attribute’s core concerns to unify understanding.

Question 4

Six parts of a quality attribute scenario -

Answer 4

• Stimulus (what triggers the system).
• Stimulus source (who/what causes the stimulus).
• Environment (situation/context the system is in when it responds).
• Artifact (system/components that react).
• Response (what system does after being triggered).
• Response measures (how we judge if the response was good).

Question 5

Name four IoT systems quality attributes -

Answer 5

Availability, performance, security and usability.

Question 6

Availability -

Answer 6

The system is ready and working when needed. It includes the ability to handle and recover from faults within a time limit. Related to:

• Security (e.g. denial-of-service attacks).
• Performance (slow response may look like failure).
• Safety (recovering from hazardous states).

Question 7

Detect faults key tactics -

Answer 7

discover when something in the system doesn’t work as intended.

• Monitor (central concept for health checking on other parts of the system).
• Heartbeat (common in real-time/IoT systems to show a process is alive).
• Ping/Echo (periodic message exchange occurs between a monitor and a process being monitored).
• Exception Detection (detection of a system condition that alters the normal flow of execution).

Question 8

Recover from faults, key tactics -

Answer 8

Tactics used to recover when a fault occurs.

• Rollback (returning to the previous working state).
• Retry (try the failed operation again).
• Graceful degradation (system continues to function, even with reduced capabilities).
• Redundant spare (use backup components).
• Exception handling (dealing with errors as they occur).
• Reconfiguration (adjusting the system to work around the fault).

Question 9

Prevent faults, key tactics -

Answer 9

• Removal from service (temporarily place a system component in an out-of-service state to mitigate potential system failure).
• Predictive model (use performance metrics to predict the onset fault).
• Exception prevention (Using methods to avoid errors (exceptions) before they happen - instead of just reacting to them.).

Question 10

Performance -

Answer 10

is about timing. Events occur, and the system must respond to them.

• Event arrival patterns: periodic, stochastic and sporadic.
• Event servicing:
  • Latency - the time taken between arrival and response.
  • Jitter - variation in latency.
  • Throughput - number of requests processed per second.

Question 11

Performance scenarios -

Answer 11

• source of stimulus (internal/external).
• Stimulus (event arrival, based on patterns).
• Artifact (system).
• Environment (normal mode; overload mode).
• Response (changes level of service, processes stimuli).
• Response measures (latency, jitter, miss rate, data loss).

Question 12

Control resource demand, tactics -

Answer 12

tactics to ensure that the system performs efficiently by controlling and managing resource usage and the demand placed on the system.

• Manage work requests (reduce the number of requests coming into the system).
• Limit event response (cap the number of events processed, policies for ignoring events).
• Prioritise events (ensure important events are processed rapidly).

Question 13

Tactics to meet performance requirements -

Answer 13

methods to improve system performance.

• Caching (store frequently used data for quick access, reducing slow data fetching).
• Load Balancing (distribute traffic across servers to optimize resources and speed up responses).
• Parallelism/Concurrency (perform multiple tasks at once to maximize resource use and reduce response time).

Question 14

Attribute-Driven design -

Answer 14

is a methodology for designing software architecture based on quality attributes (e.g., performance, scalability, security). Instead of focusing solely on functional requirements, ADD emphasises the architectural drivers that shape the system’s structure.

Question 15

Key concepts of ADD -

Answer 15

• Identify drivers: Consider quality attributes, constraints, and functional requirements.
• Decompose system: Break the architecture into subsystems and components based on drivers.
• Apply tactics/patterns: Use design patterns (e.g., microservices) to meet goals.
• Evaluate design: Assess the architecture for effectiveness.

Question 16

Why use ADD? -

Answer 16

Ensures architectural decisions align with business goals. Helps prioritise trade-offs between competing quality attributes. Provides a structured approach to designing complex systems.

Question 17

Two data Processing Paradigms

Answer 17

Real-time processing and batch processing

Question 18

Batch processing

Answer 18

• Definition: Processing data in large blocks at scheduled times.
• Tools: Apache NiFi, Apache Spark.
• Use Case: Ideal for non-time-sensitive tasks and large-scale data analytics.
• Benefits: Cost-effective for large volumes of data, efficient in resource use when real-time analysis is not required.

Question 19

Real-Time Processing

Answer 19

• Definition: Continuous input and processing of data to provide immediate output and insights.
• Tools: Apache Kafka, Apache Storm, AWS Kinesis.
• Use Case: Essential for applications requiring instant decision-making, such as fraud detection, live monitoring systems.
• Benefits: Enables timely insights and actions, critical for applications where delay can result in significant consequences

Question 20

Interoperability as quality attribute

Answer 20

Interoperability is the ability to usefully exchange information between two software components. It’s not just about sending and receiving messages, but also about ensuring that both systems understand the structure (format) and meaning (semantics) of the data — even if the components are from different manufacturers.

One key tactic:

• Use of a Message Bus - A message bus (MQTT) allows different components to publish and subscribe to messages using standardized topics.

Question 21

Modifiability as a quality attribute

Answer 21

Modifiability is the ability to make a change to the behavior of a system. These changes can occur at various stages like:

• Compile time (code changes)
• Build time (dependency changes)
• Initialization time (configuration)
• Runtime (dynamic reconfiguration or input)

One key tactic:

• Use of Configuration Parameters. Let the system’s behavior be controlled by values stored outside the code (like in files or environment variables). You can change how the system behaves without changing or recompiling the code.