What is a system execution model?
Characterizes the software’s performance in the presence of dynamic factors such as other workloads or multiple users
What does it aim to solve?
The contention for resources
Benefits of system execution model
- More precise metrics for resource contention
- Scalability for future demands
- Identification of bottlenecks
What are sources for contention?
- Multiple users
- Multiple applications
- Separate concurrent processes
- Multi-threaded to handle concurrent requests
What is server?
Represents a component of the environment that provides service to software
What is queue?
Represents jobs waiting for service
What are the main performance metrics?
- Throughput
- Utilization
- Queue length
- Residence Time
What is workload intensity?
Measure of the number of requests in given interval
What are service requirements?
The amount of time needed for the workload from each device
What is jobs-flow balance?
Completion rate is equal to the arrival rate
What are types of network model?
- Open model
- Closed model
What do we need to specify for Open QNM model?
- Workload intensity
- Service requirements
What do we need to specify for Closed QNM model?
- Number of users
- Think time
How to derive system model parameters from system model results?
Step 1: Use queue and servers to represent key computer resources
Step 2: Decide whether system should be Open QNM or closed QNM
Step 3: Determine workload intensities
Step 4: Determine the service requirements
What can’t system execution models do?
- Intricate details
- Passive resources
What are bottlenecks?
Devices with the highest utilization
-
Software Measurement and Instrumentation37
-
Performance Orientated-Design25
-
Chapter 10: Performance Patterns15
-
Chapter 2: SPE Overview12
-
Chapter 3: SPE and the UML23
-
Chapter 11: Anti-Patterns19
-
Chapter 4 : Software Execution Model14
-
Chapter 5: Web Applications and Distributed Systems14
-
Chapter 6: System Execution Models16
-
Chapter 7: Data Collection31
-
Chapter 1: Introduction8
