Tobiko Cloud: Getting Started

Tobiko Cloud is a data platform that extends SQLMesh to make it easy to manage data at scale without the waste.

We're here to make it easy to get started and feel confident that everything is working as expected. After you've completed the steps below, you'll have achieved the following:

Log in to Tobiko Cloud via the browser
Connect Tobiko Cloud to your local machine via the CLI
Connect Tobiko Cloud to your data warehouse
Verify that Tobiko Cloud interacts with your data warehouse as expected

Prerequisites

Before you start, the Tobiko team must complete a few steps.

Your Tobiko Solutions Architect will:

Set up a 1 hour meeting with you to fully onboard
Request that a new Tobiko Cloud account be created for you (single tenant by default)
Share a temporary password link that expires in 7 days
Make sure you save the password in your own password manager

To prepare for the meeting, ensure you or another attendee have data warehouse administrator rights to:

Update warehouse user and object permissions
Create new users and grant them create/update/delete permissions on a specific database (ex: database.schema.table)

For migrations from SQLMesh (open source) to Tobiko Cloud only:

Your Tobiko Solutions Architect will send you a script to extract your current state
You send that state to the Tobiko Cloud engineers to validate before the migration occurs
After validation, Tobiko Solutions Architect will schedule a migration date and meeting to move your state to Tobiko Cloud. There will be some downtime if you are running SQLMesh in a production environment.

Note: if you must be on VPN to access your data warehouse or have specific security requirements, please let us know and we can discuss options to ensure Tobiko Cloud can securely connect.

Technical Requirements:

Tobiko Cloud requires Python version between 3.9 and 3.12

Note

If you don't have a supported Python version installed, you can use uv to install it. At the time of writing, these are the suggested commands to install uv and Python:

macOS and LinuxWindows

curl -LsSf https://astral.sh/uv/install.sh | sh

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

uv python install 3.12

Log in to Tobiko Cloud

The first step to setting up Tobiko Cloud is logging in to the web interface:

We will authenticate into your Tobiko Cloud instance. If it is your first time going through this flow, your Solutions Architect will guide you on how to get SSO configured. Open the url below.
```
https://cloud.tobikodata.com/auth/login
```
Once logged in, you should see the home page. If you are not redirected, then input your Tobiko Cloud URL in the browser (ex: https://cloud.tobikodata.com/sqlmesh/tobiko/public-demo/observer/)

Your view should be empty, but the figure below shows a populated example with Tobiko Cloud running in production:

Install the `tcloud` CLI

Now we need to configure the tcloud command line interface tool.

First, open a terminal within your terminal/IDE (ex: VSCode). Then follow the following steps to install the tcloud CLI:

Create a new project directory, or an existing SQLMesh project, and navigate into it:
```
mkdir tcloud_project
cd tcloud_project
```
Create a new file called requirements.txt and add tcloud to it:
```
echo 'tcloud' > requirements.txt
```
Pypi source: tcloud

Note: your Tobiko Solutions Architect will provide you a pinned version of tcloud

Create a Python virtual environment in the project directory and install tcloud (example is using uv):

uv venv --python 3.12 --seed  # create a virtual environment inside the project directory
source .venv/bin/activate # activate the virtual environment
pip install -r requirements.txt # install the tcloud CLI
which tcloud # verify the tcloud CLI is installed in the venv in the path above

Note

You may need to run python3 or pip3 instead of python or pip, depending on your python installation.

If you do not see tcloud in the virtual environment path above, you may need to reactivate the venv:

source .venv/bin/activate
which tcloud
# expected path: /Users/person/Desktop/git_repos/tobiko-cloud-demo/.venv/bin/tcloud

Create an alias to ensure use of tcloud:

We recommend using a command line alias to ensure all sqlmesh commands run on Tobiko Cloud.

Set the alias in the terminal by running alias sqlmesh='tcloud sqlmesh' in every session.

Or add this to your shell profile file (ex: ~/.zshrc or ~/.bashrc) so you don't have to run the command every time:
```
alias sqlmesh='tcloud sqlmesh'
```
Note: the rest of the commands in this document will NOT use the alias to avoid confusion with the open source SQLMesh CLI.

Connect Tobiko Cloud to Data Warehouse

Now we're ready to connect your data warehouse to Tobiko Cloud:

Create a new file called tcloud.yaml and add the project configuration below, substituting the appropriate values for your project:

projects:
    public-demo: # TODO: update this for the project name in the URL
        url: https://cloud.tobikodata.com/sqlmesh/tobiko/public-demo/ # TODO: update for your unique URL
        gateway: tobiko_cloud
        extras: bigquery,web,github  # TODO: update bigquery for your data warehouse
default_project: public-demo # TODO: update this for the project name in the URL

If you are going through the SSO flow then, run the following command:
```
tcloud auth login
```
This will fire off the SSO flow and open a link in your browser to authenticate.

Once authenticated, you will see the following screen.

Initialize a new SQLMesh project:

tcloud sqlmesh init <your data warehouse>

Update your project's config.yaml with your data warehouse connection information:

Your new SQLMesh project will contain a configuration file named config.yaml that includes a DuckDB connection.

Replace the DuckDB connection information with your data warehouse's information.

This example shows a Bigquery warehouse connection; see more examples here.

gateways:
  tobiko_cloud: # this will use the config in tcloud.yaml for state_connection
    scheduler: # TODO: add the connection in the Tobiko Cloud Connections Page with the credentials for your data warehouse
      type: cloud

default_gateway: tobiko_cloud

model_defaults:
  dialect: bigquery # TODO: update for your data warehouse
  start: 2024-08-19 # TODO: I recommend updating this to an earlier date representing the historical data you want to backfill

# make Tobiko Cloud only allow deploying to dev environments, use env var to override in CI/CD
# allow_prod_deploy: {{ env_var('ALLOW_PROD_DEPLOY', 'false') }}

# enables synchronized deployments to prod when a pull request gets a `/deploy` command or is approved by a required approver
cicd_bot:
  type: github
  merge_method: squash
  skip_pr_backfill: false
  enable_deploy_command: true
  auto_categorize_changes:
    external: full
    python: full
    sql: full
    seed: full

# preview data for forward only models
plan:
  enable_preview: true

# list of users that are allowed to approve PRs for synchronized deployments
users:
  - username: sung_tcloud_demo
    github_username: sungchun12
    roles:
      - required_approver

Create a tcloud user in the warehouse

During your onboarding call, we will walk through instructions live to create a new tcloud data warehouse user with the necessary permissions.

SQLMesh will run as this user to create, update, and delete tables in your data warehouse. You can scope the user permissions to a specific database if needed.

Find additional data warehouse specific instructions here: Data Warehouse Integrations.
Verify the connection between Tobiko Cloud and data warehouse:

Now we're ready to verify that the connection between Tobiko Cloud and the data warehouse is working properly.

Run the info command from your terminal:
```
tcloud sqlmesh info
```
It will return output similar to this:
```
(.venv) ➜  tcloud_project git:(main) ✗ tcloud sqlmesh info
Models: 3
Macros: 0
Data warehouse connection succeeded
State backend connection succeeded
```

Verify SQLMesh functionality

Let's run a plan to verify that SQLMesh is working correctly.

Run tcloud sqlmesh plan in your terminal and enter y at the prompt to apply the changes.

tcloud sqlmesh plan

It will return output similar to this:

(.venv) ➜  tcloud_project git:(main) ✗ tcloud sqlmesh plan
======================================================================
Successfully Ran 1 tests against duckdb
----------------------------------------------------------------------
New environment `prod` will be created from `prod`
Summary of differences against `prod`:
Models:
└── Added:
    ├── sqlmesh_example.full_model
    ├── sqlmesh_example.incremental_model
    └── sqlmesh_example.seed_model
Models needing backfill (missing dates):
├── sqlmesh_example.full_model: 2024-11-24 - 2024-11-24
├── sqlmesh_example.incremental_model: 2020-01-01 - 2024-11-24
└── sqlmesh_example.seed_model: 2024-11-24 - 2024-11-24
Apply - Backfill Tables [y/n]: y
Creating physical tables ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100.0% • 3/3 • 0:00:00

All model versions have been created successfully

[1/1] sqlmesh_example.seed_model evaluated in 0.00s
[1/1] sqlmesh_example.incremental_model evaluated in 0.01s
[1/1] sqlmesh_example.full_model evaluated in 0.01s
Evaluating models ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100.0% • 3/3 • 0:00:00


All model batches have been executed successfully

Virtually Updating 'prod' ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100.0% • 0:00:00

The target environment has been updated successfully

Tobiko Cloud and SQLMesh are working!

Next steps

Your tcloud project directory should look and feel like this:

From here, if you have an existing SQLMesh project, you can copy over your existing models and macros to the models and macros directories (along with other files as needed).

You are now fully onboarded with Tobiko Cloud. We recommend reviewing the helpful links below to get familiar with SQLMesh and Tobiko Cloud.

Here's to data transformation without the waste!