A korean local news wordcloud service with over-stack (WIP)
-
Install in Local machine(Single cluster)
# Install kubernetes # Minikube guide - https://minikube.sigs.k8s.io/docs/start/ # Install helm # Helm - https://helm.sh/docs/intro/install/ # Install nodejs # n guide - https://github.com/tj/n#installation # Clone the project repository git clone https://github.com/hoongeun/krwordcloud cd ./krwordcloud/helm # helmvalues npx helmvalues build # apply to kubernetes helm install krwordcloud ./
-
Options2) Install in virtual machines(Multi cluster)
# Install Vagrant # guide - https://www.vagrantup.com/downloads # Clone the project repository git clone https://github.com/hoongeun/krwordcloud # Move to krwordcloud's vagrant directory cd ./krwordcloud/vagrant # optional - destory all staging virtualmachines vagrant destroy -f # vagrant up vagrant up
-
Options3, Will suppport) Install in native cluster(Multi cluster)
- application:
- front: react(typescript)
- backend: spring(kotlin)
- database:
- couchbase: keyword-score(proccessed data) storage
- hdfs: article storage with orc(Optimized row columnar)
- datamining: spark-operator(pyspark)
- keyword extracting algorithm: krwordrank
- datacrawler: modified korean_news_crawler
- orchestration: kubernetes, helm, vagrant
- cd: argocd
- monitoring: prometheus, grafana
- scheduler: airflow
- message queue: kafka(confluent-platform)
- hdfs 2 sink connector
- couchbase sink connector
-
Hoongeun Cho (me@hoongeun.com): Architect, Project management, orchestration, application, spark
-
Jinseong Choi: airflow