tickets-api

README

tickets-api

Tickets API Simulation

Seed the Database

• Install dependencies
  • pip install -r requirements.txt
• Update the config/config.json with the DB information
• Execute the Python script to generate Users and Events
  • Secure CRDB: python db-seed/seed.py --secure --config config/config.json --num_users 10000 --num_events 10000 --num_purchases 1 --num_cancellations 1 --num_payments 1
  • Insecure CRDB: python db-seed/seed.py --config config/config.json --num_users 10000 --num_events 10000 --num_purchases 1 --num_cancellations 1 --num_payments 1

We only generated 1 row for the rest of the tables, this is just to trigger their creation and have a sample record. To generate a large payload of a user buying tickets for events and seed the purchases and cancellations tables, use the following SQL in Cockroach:

INSERT INTO purchases (id, user_id, event_id, status)
 SELECT 
 	gen_random_uuid() as id,
    u.id AS user_id,
    e.id AS event_id,
    CASE WHEN random() < 0.5 THEN 'cancelled' ELSE 'confirmed' END AS status
FROM 
    (SELECT id FROM users LIMIT 100) AS u
CROSS JOIN 
    (SELECT id FROM events ORDER BY random() LIMIT (random() * 450 + 50)::int) AS e;

This query will generate a large payload of purchases per user (~450 rows).

Tickets API Endpoints

• GET: /user/:uuid/purchases - get user purchases
• GET: /user/:uuid/purchases/cancellations - get user cancellations
• GET: /search/user/:uuid - search users by ID http://localhost:3001/search/user/27912e48-88f5-4dca-b549-88da7bdda1f4 for go-pg
• GET: /search/user/:uuid - search users by ID http://localhost:3002/search/user/27912e48-88f5-4dca-b549-88da7bdda1f4 for pgx

Start GO-PG Tickets API Endpoint

cd gopg-api

go mod init gopg-api
go mod tidy
go run *.go

Collect Test Data Points for the k6 Test

SELECT purchase.user_id
  FROM purchases AS purchase LEFT
    JOIN users AS users
     ON users.id = purchase.user_id LEFT
    JOIN events AS events
     ON events.id = purchase.event_id order by random() limit 100 ;

The results will need to be added to the k6 const names = []

Run K6 Stress Test

k6 run k6-gopg.js

Start PGX Tickets API Endpoint

cd pgx-api

go mod init pgx-api
go mod tidy
go run *.go

Run K6 Stress Test

k6 run k6-pgx.js

Implicit/Explicit Transaction Example

• This example focuses on querying the users table
• This example requires Jaeger (can be deployed in a docker container) to trace the API execution and it's SQL statement basic metrics
• This example is built to test Read Committed transactions in CockroachDB 23.2.x (beta at the moment of this writing)

cd read-commit

go mod init implicit-explicit
go mod tidy
go run *.go

NOTE: for K6 Stress test, the ramp up in this examples have been set to 1m.

Test Implicit Transactions

• Slightly modify your K6 script to hard code some user UUIDs in an array
  • const uuids = [...uuids...]
• API endpoint: http://localhost:8080/implicit/users/$uuid
• Run the stress test
  • k6 run --vus 500 --duration 5m k6-implicit.js

Test Explicit Transactions (v23.2.x+)

• Slightly modify your K6 script to hard code some user UUIDs in an array
  • const uuids = [...uuids...]
• API endpoint: http://localhost:8080/explicit/users/$uuid
• Enable Read Commit in CockroachDB
  • root@192.168.86.74:26257/defaultdb ?> SET CLUSTER SETTING sql.txn.read_committed_syntax.enabled = 'true';
  • At the session level set: set default_transaction_isolation = 'read committed';
• Run the stress test
  • k6 run --vus 500 --duration 5m k6-explicit.js