VeloDB Cloud Quick Start

Estimated completion time: 20 minutes

This quick start shows you how to create a VeloDB Cloud warehouse, load the SSB-Flat dataset, and run analytical queries.

Step 1. Create your account

You need to use a business email address to sign up. Personal emails are not supported. No credit card is required. The free trial gives you US$300 in credits and 30 days to explore VeloDB Cloud.

To create a new account:

Go to the VeloDB Cloud console and then click Sign up.
Choose one of the following sign-up methods and complete the registration.
- Google: single sign-on with your Google account
- Microsoft: single sign-on with your Microsoft account
- Email: traditional business email and password registration

Step 2. Create a warehouse

To create a warehouse:

Log in to the VeloDB Cloud console.
In the upper-left corner, click Create New Warehouse.
In the SaaS area, click Create Warehouse.
Enter the following information:
- Warehouse Name: enter velodb-quickstart.
- Cloud Platform: select AWS.
- Region: choose a region close to you for lowest latency. The following regions are available for the sample dataset:
  - us-east-1 (US East, N. Virginia)
  - us-west-1 (US West, N. California)
  - us-west-2 (US West, Oregon)
  - eu-central-1 (Europe, Frankfurt)
  - eu-west-1 (Europe, Ireland)
  - ap-southeast-1 (Asia Pacific, Singapore)
  - ap-northeast-1 (Asia Pacific, Tokyo)
  - me-south-1 (Middle East, Bahrain)
You can keep the default cluster specification (4 vCPU, 32 GB RAM, and 200 GB cache) and advanced settings for this tutorial. Click Get started to provision the warehouse.

It usually takes 3 to 5 minutes to provision a warehouse. Continue when the warehouse status changes to Running.

Step 3. Connect to your warehouse

It is recommended that you use the built-in SQL Editor to connect to your warehouse. For other connection methods, see Connecting by MySQL Protocol.

To connect to your warehouse:

In the left navigation pane, click SQL Editor.
Enter your warehouse password and then click Connect.

Run the following statement to verify the connection:

SHOW DATABASES;

You can see at least the following system database:

+--------------------+
| Database           |
+--------------------+
| information_schema |
+--------------------+

On the SQL Editor home page, you can explore the following use case demos and SQL templates to see how VeloDB Cloud works and learn how common data analysis tasks are performed with SQL.

Use case demos
- Real-time analytics: track product activity and power live dashboards in real time.
- Observability: search and troubleshoot logs, metrics, and traces at scale.
SQL templates
- Complex Joins: learn various join queries step by step: INNER, LEFT, RIGHT, FULL, CROSS, LEFT SEMI, and LEFT ANTI.
- Full Text Search: search documents and logs using text indexes.
- Semi-Structured Data: store and query flexible JSON schemas without compromising SQL performance.

Step 4. Preview the SSB-Flat benchmark dataset from Amazon S3

The public SSB-Flat SF1 dataset contains about 6 million rows in a denormalized table. You can preview the file, create a table, load the Parquet file, and verify the row count. You do not need Amazon S3 credentials because the sample dataset is publicly accessible.

To preview the dataset from Amazon S3:

Replace every occurrence of <region> in the following statement with the AWS region where your warehouse is located, such as us-east-1.

-- Replace <region> with your warehouse AWS region.
-- The bucket name, endpoint, and region must use the same value.
SELECT
    lo_orderkey,
    lo_orderdate,
    c_name,
    c_nation,
    lo_revenue,
    p_brand
FROM S3(
    "uri" = "s3://velodb-import-data-<region>/ssb-flat-sf1/ssb_flat_001.parquet",
    "format" = "parquet",
    "s3.endpoint" = "s3.<region>.amazonaws.com",
    "s3.region" = "<region>"
)
LIMIT 5;

Run the statement in the SQL Editor.

Expected output

lo_orderkey	lo_orderdate	c_name	c_nation	lo_revenue	p_brand
1	1996-08-30	Customer#000003506	INDIA	41169.9	MFGR#327
1	1998-12-27	Customer#000020886	KENYA	43440.7	MFGR#217
1	1994-11-01	Customer#000011876	ROMANIA	8084.1	MFGR#510
1	1995-05-28	Customer#000012832	KENYA	34849.0	MFGR#531
1	1996-08-05	Customer#000008486	INDONESIA	36930.1	MFGR#409

Step 5. Create the table and load data

Run the following script in the SQL Editor in order. Replace <region> with the region where your warehouse is located.

View the SQL statements

-- 1. Create a database for the tutorial.
CREATE DATABASE IF NOT EXISTS ssb;
USE ssb;

-- 2. Create a wide table that matches the Parquet file schema.
CREATE TABLE IF NOT EXISTS ssb_flat (
    lo_orderkey        BIGINT NOT NULL,
    lo_linenumber      BIGINT NOT NULL,
    lo_custkey         BIGINT NOT NULL,
    lo_partkey         BIGINT NOT NULL,
    lo_suppkey         BIGINT NOT NULL,
    lo_orderdate       DATE NOT NULL,
    lo_commitdate      DATE NOT NULL,
    lo_orderpriority   VARCHAR(15) NOT NULL,
    lo_shippriority    BIGINT NOT NULL,
    lo_shipmode        VARCHAR(10) NOT NULL,
    lo_year            INT NOT NULL,
    lo_month           INT NOT NULL,
    lo_weeknum         INT NOT NULL,
    d_datekey          BIGINT NOT NULL,
    d_dayofweek        VARCHAR(10) NOT NULL,
    d_month            VARCHAR(10) NOT NULL,
    d_yearmonth        VARCHAR(10) NOT NULL,
    lo_quantity        BIGINT NOT NULL,
    lo_extendedprice   DOUBLE NOT NULL,
    lo_discount        DOUBLE NOT NULL,
    lo_revenue         DOUBLE NOT NULL,
    lo_supplycost      DOUBLE NOT NULL,
    lo_tax             DOUBLE NOT NULL,
    c_custkey          BIGINT NOT NULL,
    c_name             VARCHAR(25) NOT NULL,
    c_nation           VARCHAR(15) NOT NULL,
    c_region           VARCHAR(12) NOT NULL,
    c_city             VARCHAR(10) NOT NULL,
    c_mktsegment       VARCHAR(10) NOT NULL,
    s_suppkey          BIGINT NOT NULL,
    s_name             VARCHAR(25) NOT NULL,
    s_nation           VARCHAR(15) NOT NULL,
    s_region           VARCHAR(12) NOT NULL,
    s_city             VARCHAR(10) NOT NULL,
    p_partkey          BIGINT NOT NULL,
    p_name             VARCHAR(22) NOT NULL,
    p_brand            VARCHAR(9) NOT NULL,
    p_category         VARCHAR(7) NOT NULL,
    p_mfgr             VARCHAR(6) NOT NULL,
    p_color            VARCHAR(11) NOT NULL,
    p_type             VARCHAR(25) NOT NULL,
    p_size             BIGINT NOT NULL,
    p_container        VARCHAR(10) NOT NULL,
    INDEX idx_p_type (p_type) USING INVERTED PROPERTIES("parser" = "english")
)
DUPLICATE KEY(lo_orderkey)
DISTRIBUTED BY HASH(lo_orderkey) BUCKETS 48;

-- 3. Load the Parquet file from the public sample bucket.
-- Replace <region> in the bucket name, endpoint, and region option.
INSERT INTO ssb.ssb_flat
SELECT *
FROM S3(
    "uri" = "s3://velodb-import-data-<region>/ssb-flat-sf1/*.parquet",
    "format" = "parquet",
    "s3.endpoint" = "s3.<region>.amazonaws.com",
    "s3.region" = "<region>"
);

-- 4. Verify that the data was loaded.
SELECT COUNT(*) AS total_rows FROM ssb.ssb_flat;

The load usually takes about 30 seconds to 1 minute. The expected output is as follows:

+------------+
| total_rows |
+------------+
|    6000000 |
+------------+

Step 6. Run analytical queries

The following queries use the same dataset to demonstrate common analytical workloads. Expand a query to copy and run it in the SQL Editor.