DZone
Data Engineering
Data
Generating Simulated Streaming Data

Generating Simulated Streaming Data

In this article, learn more about using the Python library, Faker, to build synthetic data for tests and utilize Pulsar to send messages to topics at scale.

👁 Tim Spann user avatar

Tim Spann

👁 DZone Core
CORE ·

👁 Timothy Spann user avatar

Timothy Spann

Mar. 06, 22 · Tutorial

Likes (5)

Comment

Save

8.9K Views

Join the DZone community and get the full member experience.

Join For Free

For demos, system tests, and other purposes, it is good to have a way to easily produce realistic data at scale utilizing a schema of our own choice.

Fortunately, there is a great library for Python called Faker that lets us build synthetic data for tests. With a simple loop and a Pulsar produce call, we can send messages to topics at scale.

pip3 install Faker

Let's build a topic to send data to as below:

bin/pulsar-admin topics create persistent://public/default/fakeuser

Step 1:

We build a Faker record by adding providers to the Faker object.

Python

from faker import Faker
from faker.providers import internet, address, automotive, barcode, company, date_time, geo, job, misc, person
from faker.providers import phone_number, user_agent

fake = Faker()
fake.add_provider(internet)

Step 2:

Create or Import Your Schema to model your data and have consistent data in your Pulsar topic for consumers to consume.

Python

# Pulsar Message Schema
class PulsarUser (Record):
 created_dt = String()
 user_id = String()
 ipv4_public = String()
 email = String()
 user_name = String()
 cluster_name = String()
 city = String()
 country = String()
 postcode = String()
 street_address = String()
 license_plate = String()
 ean13 = String()
 response = String()
 comment = String()
 company = String()
 latitude = Float()
 longitude = Float()
 job = String()
 email_me = Boolean()
 secret_code = String()
 password = String()
 first_name = String()
 last_name = String()
 phone_number = String()
    user_agent = String()

Step 3:

Connect to your cluster and build a producer.

Python

client = pulsar.Client('pulsar://pulsar1:6650')
producer = client.create_producer(topic='persistent://public/default/fakeuser',
 schema=JsonSchema(PulsarUser),
                                  properties={"producer-name": "fake-py-sensor","producer-id": "fake-user" })

Step 4:

Create records (you can do this in a finite or infinite loop).

Python

userRecord = PulsarUser()
uuid_key = '{0}_{1}'.format(strftime("%Y%m%d%H%M%S",gmtime()),uuid.uuid4())
userRecord.created_dt = fake.date() 
userRecord.user_id = uuid_key
userRecord.ipv4_public = fake.ipv4_public()

Step 5:

Send record:

Python

producer.send(userRecord,partition_key=str(uuid_key))

An example row of produced JSON data is as follows:

JSON

{
 'created_dt': '1974-01-07', 
 'user_id': '20220304192446_045a2724-6f9e-4968-ae19-a5a1a095e57b', 
 'ipv4_public': '207.116.194.88', 
 'email': '[email protected]', 
 'user_name': 'qpearson', 
 'cluster_name': 'memory-story-see', 
 'city': 'Elizabethview', 
 'country': 'Mauritius', 
 'postcode': '01045', 
 'street_address': '352 Rodriguez Rue', 
 'license_plate': '6-79707I', 
 'ean13': '3151191404713', 
 'response': 'Quality-focused logistical conglomeration', 
 'comment': 'implement value-added relationships', 
 'company': 'Harper LLC', 
 'latitude': 88.5392145, 'longitude': -7.466258, 
 'job': 'Development worker, community', 
 'email_me': None,
 'secret_code': '28b5519f4bb38cd9fc52aa9bb7bca1aa', 
 'password': '+5u&%TTkHt', 
 'first_name': 'Johnny', 
 'last_name': 'Hoffman', 
 'phone_number': '9669534677', 
 'user_agent': 'Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 6.2; Trident/3.1)'
}

We can check that our schema is loaded for the topic with the following command.

bin/pulsar-admin schemas get persistent://public/default/fakeuser

We can consume the test messagges with a command line consumer.

bin/pulsar-client consume "persistent://public/default/fakeuser" -s fakeuser-consumer -n 0

We can query the topic like a table via Pulsar SQL.

Source

https://github.com/tspannhw/FLiP-PY-FakeDataPulsar

Data (computing)

Opinions expressed by DZone contributors are their own.

Cutting Data Pipeline Costs and Data Freshness Issues With Netflix Maestro and Apache Iceberg: A Practical Tutorial
From ETL to Lakeflow: Shifting to a Declarative Data Paradigm
Stop Loading Everything into Redshift: A Spectrum + Iceberg Pattern for Hybrid Analytics
Operationalizing Enterprise AI at Scale: Architecture, Governance, and Adoption

URL: https://dzone.com/articles/generating-simulated-streaming-data