ORIS Predictive Autoscaler

Project Structure

.
├── inter-arrival-collector # Service that extarct CDF from inter arrival times
│   ├── collect_inter_arrivals.py
│   ├── Dockerfile
│   └── requirements.txt
├── k6 # Implementation of the test workload
│   ├── build-and-run.sh
│   ├── Dockerfile
│   ├── lib
│   │   ├── index.js
│   │   ├── LICENSE.txt
│   │   ├── package.json
│   │   ├── probability-distributions-k6
│   │   ├── README.md
│   │   └── test
│   │       └── index.js
│   └── rabbitmq-test.js
├── k8s # Various configuration for used services
│   ├── grafana.yaml
│   ├── inter-arrival-collector.yaml
│   ├── kafdrop.yaml
│   ├── kafka.yaml
│   ├── kube-state-metrics-rbac.yaml
│   ├── kube-state-metrics.yaml
│   ├── namespace.yaml
│   ├── prometheus.yaml
│   ├── rabbitmq-config.yaml
│   ├── rabbitmq.yaml
│   ├── service.yaml
│   ├── sirio-controller-rbac.yaml
│   └── sirio-controller.yaml
├── service # Consumer Service
│   ├── Dockerfile
│   ├── requirements.txt
│   └── service.py
├── sirio-controller # Implementation of a Horizontal Scaler using Sirio projection of Rejection Rate
│   ├── Dockerfile
│   ├── pom.xml
│   └── src
│       ├── main
│       └── test
└── start.sh

Quick Start

0. Get all the Code

The command for coping the repository.

git clone https://github.com/Berags/oris-predictive-autoscaler.git

1. Cluster Deployment

First you should deploy the sistem using the following script.

# make the script executable
chmod +x deploy.sh

# start the cluster
./start.sh

At the end it will be exposed some Web GUI to monitor the system.

2. Run k6 tests

After the system deployment it spossible to subject it with some workload with the following code:

# make the script executable
chmod +x k6/build-and-run.sh

# start the cluster
./k6/build-and-run.sh

This is instrumented to show some information about the workload.

3. Exposed Services

• RabbitMQ Management: http://localhost:15672
  • Username: admin
  • Password: password
• Grafana Dashboard: http://localhost:3000
  • Username: admin
  • Password: admin
• Prometheus Web GUI: http://localhost:9090
• Kafka Web GUI: http://localhost:9000

Configuration

Python Service

The Python service can be configured using environment variables:

• RABBITMQ_HOST: RabbitMQ Host (default: rabbitmq-service)
• RABBITMQ_PORT: RabbitMQ Port (default: 5672)
• RABBITMQ_USER: RabbitMQ Username (default: admin)
• RABBITMQ_PASS: RabbitMQ Password (default: password)

RabbitMQ

RabbitMQ is configured with:

• Management interface enabled
• Porta AMQP: 5672
• Porta Management: 15672
• Default Credentials: admin/password