benchy

hamza ali taş

a command-line application for benchmarking a specific TimescaleDB SQL query.

requirements and setup

requires a PostgreSQL instance to connect

• if you have docker installed
  • run setup_db.sh
  • creates interview network
  • creates timescaledb container named interview-db with user postgres and password 12345 and attached to interview network
  • exposes 5432 port to host machine
  • creates the database and applies the schema
  • imports test data into the container database

build and run

golang:

• go install then run benchy
• go build then run ./benchy

docker:

• build: docker build -t benchy .
• run: data/query_params.csv | docker run --rm [--network interview] -i benchy
• or just run script ./run.sh

usage

examples

# golang
benchy -config config.json -file data/query_params.csv
cat data/query_params.csv | benchy -config config.json -host some.ip.val.ue

# docker
cat data/query_params.csv | docker run --rm -i benchy
cat data/query_params.csv | docker run --rm -i benchy -port 1234 

possible configuration flags as follows;

-config string
    custom config path
-db string
    database schema name (default "postgres")
-file string
    csv file input path for query parameters
-host string
    database host (default "localhost")
-password string
    database password
-port int
    database port (default 5432)
-ssl
    database ssl mode
-user string
    database user (default "postgres")
-worker int
    worker count (default 5)

---

notes

• if the -file flag not provided, program reads STDIN as the csv file input
• if the -file flag provided, program reads the given file path and ignores STDIN input
• configuration can be passed by providing a config.json file as flag, however, flags will always override the configuration

---

example json configuration

{
  "worker_count": 5,
  "postgres": {
    "host":"interview-db",
    "port": 5432,
    "ssl": false,
    "db": "homework",
    "user": "postgres",
    "password": "12345"
  }
}

implementation

project scaffold with go mod init, built using go standard libraries, and the only external dependency is the pq postgresql driver.

concurrency provided by using go routines communication with channels and synchronized using wait groups.

program splits into four main parts;

• read config
• start workers
• read csv line by line and parse assign jobs to workers
• collect results and print

data flow diagram;