For a project, i have to compare between PostgreSQL and Elasticsearch databases for full text searches.
I have to setup a database cluster for a website supporting more than 5000 users simultaneously.
I'm not an database administrator expert so i've to Googling a lot.

After some reading i've read that PostgreSQL and ElasticSearch are good full text searches solutions but i've not found which solution is best adapted for my needs.
I will update this Git as soon as i'm getting better results.
If you have any suggestions or remarks/questions, do not hesitate to submit an issue.

• I'm mostly using Docker Swarm to set up a cluster.
• The database requests are mostly read requests.
• The dataset to benchmark the databases comes from IMDB actors names (3 960 899 rows).
• The benchmark app (a simple PHP script) queries the database endlessly and display the numbers queries/s performed.

• Docker 1.13 and Docker-compose
• A good machine whith enough CPU and RAM
• Curl
• Extract the dataset tar zxvf datasets.tar.gz

docker-compose -f ./postgresql/postgresql.yml up -d postgresql

echo "Importing the database, please wait..." && sleep 5 && ./postgresql/psql_importdb.sh

docker-compose -f ./postgresql/postgresql.yml up -d benchmark_app

docker-compose -f ./postgresql/postgresql.yml logs benchmark_app

docker-compose -f ./postgresql/postgresql.yml kill ; \
docker-compose -f ./postgresql/postgresql.yml rm ; \
rm -rf ./postgresql/postgres_db/*

You must set Docker as an Docker Cluster : docker swarm init

echo "127.0.0.1       elasticsearch.lan" >> /etc/hosts

sysctl -w vm.max_map_count=262144

docker stack deploy -c ./elasticsearch/elasticsearch_swarm.yml benchmark

echo "Importing the database in the background. This can take a while..." && ./elasticsearch/import_db.sh

./benchmark_app/stress.sh

docker service rm $(docker service ls -q)

The benchmark scripts use one single CPU core. To benefit to all yours cores, you can scale the container

docker-compose -f ./postgresql/postgresql.yml scale benchmark_app=4

soon

Soon