https://github.com/dmarrazzo/bo-training
Red Hat Business Optimizer trainings lab 7.2.1
https://github.com/dmarrazzo/bo-training
Last synced: 3 months ago
JSON representation
Red Hat Business Optimizer trainings lab 7.2.1
- Host: GitHub
- URL: https://github.com/dmarrazzo/bo-training
- Owner: dmarrazzo
- Created: 2019-02-15T08:02:55.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2019-02-16T09:51:30.000Z (over 6 years ago)
- Last Synced: 2025-01-22T20:51:20.091Z (5 months ago)
- Language: Java
- Size: 793 KB
- Stars: 1
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Business Optimizer Training labs
## lab01
### Goal
Practice the annotations
### Instruction
When you find the planning entity, add the property `difficultyComparatorClass` that is used to initialize the heuristic construction.
@PlanningEntity(difficultyComparatorClass = CloudProcessDifficultyComparator.class)
When you find the planning variable, add the property `strengthComparatorClass` that is used to initialize the heuristic construction.
@PlanningVariable(valueRangeProviderRefs = {"..."},
strengthComparatorClass = CloudComputerStrengthComparator.class)
Spot the planning solution and add:
@PlanningSolution
Other annotations for the solution:
@PlanningScore
@ValueRangeProvider(id = "computerRange")
@ProblemFactCollectionProperty
@PlanningEntityCollectionProperty
### Solution
Check lab02 model that has all the annotations
## lab02
### Goal
Practices constriaints defintitions with DRL
### Instruction
Open `lab02/src/main/resources/org/optaplanner/examples/cloudbalancing/solver/cloudBalancingScoreRules.drl`
Add the following rules:
- No more than 4 processes per computer
- Distribute network bandwidth fairly
Hints:
- Write one rule per time and test it
- Before implementing the 3rd rule read in the documentation **Fairness Score Constraints** in the **score trap** section
### Solution
#### No more than 4 processes per computer
```js
rule "atMost4ProcessesPerComputer"
when
$computer : CloudComputer()
accumulate(
$c : CloudProcess(computer == $computer);
$count : count($c);
$count > 4
)
then
scoreHolder.addHardConstraintMatch(kcontext, 4 - $count.intValue());
end
```
#### Pitfalls
A computer that has 6 processes is worse than a computer that has 5 processes.
The score function should reflect that, even if the customer's business analyst does not want to think
about any solution for which a computer has more than 4 computers (because it's an infeasible solution).
This code might trigger a http://docs.optaplanner.org/latest/optaplanner-docs/html_single/index.html#scoreTrap
```js
rule "atMost4ProcessesPerComputer"
when
...
then
scoreHolder.addHardConstraintMatch(kcontext, -1); // BAD: score trap
end
```
#### Distribute network bandwidth fairly
```js
rule "distributeNetworkBandwidthFairly"
when
$computer : CloudComputer()
accumulate(
CloudProcess(
computer == $computer,
$requiredNetworkBandwidth : requiredNetworkBandwidth);
$total : sum($requiredNetworkBandwidth)
)
then
scoreHolder.addSoftConstraintMatch(kcontext, - ($total * $total));
end
```
## lab03
### Goal
Practice the benchmark
### Instruction
Open: `lab03/src/main/resources/logback.xml`:
- change level to `info`
Work in lab03 project where you will find a clean project.
Search the benchmark class: `CloudBalancingBenchmarkHelloWorld`.
Open the benchmark configuration file used in the:
`lab03/src/main/resources/org/optaplanner/examples/cloudbalancing/optional/benchmark/cloudBalancingBenchmarkConfig.xml`
There are 4 sections:
- `problemBenchmarks` data sets used for the benchmark
- `solver` here it's possible to reduce the termination time
- `solverBenchmark` 5 different solver configuration
Run the benchmarker reducing the termination time to 2 minutes:
- `(2 minutes * 3 configuration * 5 data sets) / parallel threads`, in a laptop with 4 cores, leads to 17 minutes of wait time
To get a more predictable result, configure the number of thread to fixed value.
Which is the property to set it up?
Explore the report.
Optionally, try the other configurations available.