ITEC 3030 TERMS I HAVE TO REMEMBER

Single point of failure

One part whose failure stops the entire system from working.

Cascading failure

One failure triggers a chain of failures through dependencies or load shifts.

Workload

tasks

throughput

tasks handled/time

response time

wait per task; higher throughput lowers wait if workload fixed.

The knee

Load point where performance worsens rapidly; small load increases cause big response time jumps.

Partition and Parallelize

Split work into independent parts and process them simultaneously for speed.

Reduce Contention via Replication

Duplicate resources so users share load, reducing bottlenecks.

Scaling up (vertical)

Add power to one machine (CPU, RAM) to handle more.

Scaling out (horizontal)

Add more machines/nodes to share workload in parallel.

Graceful degradation under overload

Reduce service to protect current users instead of failing completely.

Load balancing

Distribute incoming work evenly across multiple resources.

Caching

Store recent results for faster repeated access.

Fault tolerance by redundancy

Use backup components to continue service after a failure.

ADP

No dependency loops; components should not form cycles in dependency graphs.

SDP

Depend on more stable components to prevent frequent changes affecting others.

Connectors

Infrastructure that links components for interaction, without performing domain tasks.

MVC

Controller handles input, updates Model; Model notifies View to update display.

Functional scenario

Describes step-by-step system actions for a specific task.

Quality scenario

Tests system’s performance or behavior under specific conditions.

Scenario

Story to test architecture against requirements

Good scenario

Clear, realistic, specific, and measurable for success.