TPC-C Web Server

A webserver based on the TPC-C database benchmark. It consists of 5 transaction types interacting with a complex database mimicking a real-life use case.

The webserver provided routes are:

Path	Method	Body
/init_db	POST	`{ "warehouses": 10 }` Optional. Defaults to WAREHOUSES.
/orders/	POST	`{ "w_id": 1, "c_id": 1 }`
/payment/	POST	`{ "w_id": 1, "c_id": 1 }`
/customers/{customer_id}/orders	POST
/warehouses/{warehouse_id}/deliveries	POST
/warehouses/{warehouse_id}/stock	GET

Notes:

All routes are POST requests
w_id is the warehouse id, c_id is the customer id

How to run

Docker - SQLite

docker compose -f deploy/sqlite.compose.yml up -d

Docker - Postgres

docker compose -f deploy/postgres.compose.yml up -d

This builds and starts the webserver along with a single Postgres database. Please check deploy/postgres.compose.yml for configuration details such as username, password, db name for postgres and gunicorn config for server.

Local

# Init
python -m virtualenv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Set `DATABASE_URL` and `WAREHOUSES` environment variable as required.
export DATABASE_URL=postgresql://localhost/bench_sa
export WAREHOUSES=10 # Default: 10 

# Run
cd src/ && gunicorn --bind=localhost:5000 app:app

Init Database

Initialise database with all tables according to TPC-C schema and fill up with seed data for warehouses, districts, customers, etc.

curl --request POST http://localhost:5000/init_db

Optionally, you can also pass the amount of warehouses,

curl --request POST 'http://localhost:5000/init_db' \
    --header 'Content-Type: application/json' \
    --data '{
        "warehouses": 20
    }'

LiteFS

./litefs/run.sh up # Docker up all
./litefs/run.sh down # Docker down all

# For more info
./litefs/run.sh help

Experiment

Duration: 30 minutes

Virtual Users: 2

5 Cloudlab instances with following specifications:


CPU	2 x Xeon E5-2660 v3 processors (10 cores each, 2.6Ghz or more)
RAM	256GB Memory (16 x 16GB DDR4 DIMMs)
Disks	1 x 900GB 10K SAS Drive
NIC	1GbE Quad port embedded NIC (Intel)
NIC	1 x Solarflare Dual port SFC9120 10G Ethernet NIC

Each node had the following processes running:

Node	Process
0	nginx
1	primary
2	replica
3	replica
4	benchmark client

Results

MQTh (Throughput of Orders Transaction) = 11.28 requests / second

Transaction	P50	P90	P95	P99
order	0.0693	0.1102	0.1236	0.3309
payment	0.044	0.081	0.0912	0.3046
delivery	0.0945	0.136	0.152	0.3888
order_status	0.0275	0.0485	0.0672	0.0951
stock_level	0.4376	0.5132	0.5352	0.5705

Acknowledgements

TPC-C code is based on Python ORM Benchmark.

fadhilkurnia/tpcc-web