Skip to main content

dbt Developer Blog

Technical tutorials from the dbt Community.

Start here

· 11 min read

Before I delve into what makes this particular solution "intelligent", let me back up and introduce CI, or continuous integration. CI is a software development practice that ensures we automatically test our code prior to merging into another branch. The idea being that we can mitigate the times when something bad happens in production, which is something that I'm sure we can all resonate with!

· 8 min read

Testing the quality of data in your warehouse is an important aspect in any mature data pipeline. One of the biggest blockers for developing a successful data quality pipeline is aggregating test failures and successes in an informational and actionable way. However, ensuring actionability can be challenging. If ignored, test failures can clog up a pipeline and create unactionable noise, rendering your testing infrastructure ineffective.

· 7 min read

Imagine you were responsible for monitoring the safety of a subway system. Where would you begin? Most likely, you'd start by thinking about the key risks like collision or derailment, contemplate what causal factors like scheduling software and track conditions might contribute to bad outcomes, and institute processes and metrics to detect if those situations arose. What you wouldn't do is blindly apply irrelevant industry standards like testing for problems with the landing gear (great for planes, irrelevant for trains) or obsessively worry about low probability events like accidental teleportation before you'd locked down the fundamentals. 

When thinking about real-world scenarios, we're naturally inclined to think about key risks and mechanistic causes. However, in the more abstract world of data, many of our data tests often gravitate towards one of two extremes: applying rote out-of-the-box tests (nulls, PK-FK relationships, etc.) from the world of traditional database management or playing with exciting new toys that promise to catch our wildest errors with anomaly detection and artificial intelligence. 

Between these two extremes lies a gap where human intelligence goes. Analytics engineers can create more effective tests by embedding their understanding of how the data was created, and especially how this data can go awry (a topic I've written about previously). While such expressive tests will be unique to our domain, modest tweaks to our mindset can help us implement them with our standard tools. This post demonstrates how the simple act of conducting tests by group can expand the universe of possible tests, boost the sensitivity of the existing suite, and help keep our data "on track". This feature is now available in dbt-utils.

· 9 min read

In seventh grade, I decided it was time to pick a realistic career to work toward, and since I had an accountant in my life who I really looked up to, that is what I chose. Around ten years later, I finished my accounting degree with a minor in business information systems (a fancy way of saying I coded in C# for four or five classes). I passed my CPA exams quickly and became a CPA as soon as I hit the two-year experience requirement. I spent my first few years at a small firm completing tax returns but I didn't feel like I was learning enough, so I went to a larger firm right before the pandemic started. The factors that brought me to the point of changing industries are numerous, but I’ll try to keep it concise: the tax industry relies on underpaying its workers to maintain margins and prevent itself from being top-heavy, my future work as a manager was unappealing to me, and my work was headed in a direction I wasn’t excited about.

· 7 min read

Why we built this: A brief history of the dbt Labs Professional Services team

If you attended Coalesce 2022, you’ll know that the secret is out — the dbt Labs Professional Services team is not just a group of experienced data consultants; we’re also an intergalactic group of aliens traveling the Milky Way on a mission to enable analytics engineers to successfully adopt and manage dbt throughout the galaxy.

· 11 min read

Once your data warehouse is built out, the vast majority of your data will have come from other SaaS tools, internal databases, or customer data platforms (CDPs). But there’s another unsung hero of the analytics engineering toolkit: the humble spreadsheet.

Spreadsheets are the Swiss army knife of data processing. They can add extra context to otherwise inscrutable application identifiers, be the only source of truth for bespoke processes from other divisions of the business, or act as the translation layer between two otherwise incompatible tools.

Because of spreadsheets’ importance as the glue between many business processes, there are different tools to load them into your data warehouse and each one has its own pros and cons, depending on your specific use case.

· 6 min read

Data is an industry of sidesteppers. Most folks in the field stumble into it, look around, and if they like what they see, they’ll build a career here. This is particularly true in the analytics engineering space. Every AE I’ve talked to had envisioned themselves doing something different before finding this work in a moment of serendipity. This raises the question, how can someone become an analytics engineer intentionally? This is the question dbt Labs’ Foundry Program aims to address.

· 13 min read

For years working in data and analytics engineering roles, I treasured the daily camaraderie sharing a small office space with talented folks using a range of tools - from analysts using SQL and Excel to data scientists working in Python. I always sensed that there was so much we could work on in collaboration with each other - but siloed data and tooling made this much more difficult. The diversity of our tools and languages made the potential for collaboration all the more interesting, since we could have folks with different areas of expertise each bringing their unique spin to the project. But logistically, it just couldn’t be done in a scalable way.

So I couldn’t be more excited about dbt’s polyglot capabilities arriving in dbt Core 1.3. This release brings Python dataframe libraries that are crucial to data scientists and enables general-purpose Python but still uses a shared database for reading and writing data sets. Analytics engineers and data scientists are stronger together, and I can’t wait to work side-by-side in the same repo with all my data scientist friends.

Going polyglot is a major next step in the journey of dbt Core. While it expands possibilities, we also recognize the potential for confusion. When combined in an intentional manner, SQL, dataframes, and Python are also stronger together. Polyglot dbt allows informed practitioners to choose the language that best fits your use case.

In this post, we’ll give you your hands-on experience and seed your imagination with potential applications. We’ll walk you through a demo that showcases string parsing - one simple way that Python can be folded into a dbt project.

We’ll also give you the intellectual resources to compare/contrast:

  • different dataframe implementations within different data platforms
  • dataframes vs. SQL

Finally, we’ll share “gotchas” and best practices we’ve learned so far and invite you to participate in discovering the answers to outstanding questions we are still curious about ourselves.

Based on our early experiences, we recommend that you:

Do: Use Python when it is better suited for the job – model training, using predictive models, matrix operations, exploratory data analysis (EDA), Python packages that can assist with complex transformations, and select other cases where Python is a more natural fit for the problem you are trying to solve.

Don’t: Use Python where the solution in SQL is just as direct. Although a pure Python dbt project is possible, we’d expect the most impactful projects to be a mixture of SQL and Python.

· 21 min read

IMPORTANT: This document serves as the temporary location for information on how to design and structure your metrics. It is our intention to take this content and turn it into a Guide, like How we structure our dbt projects, but we feel that codifying information in a Guide first requires that metrics be rigorously tested by the community so that best practices can arise. This document contains our early attempts to create best practices. In other words, read these as suggestions for a new paradigm and share in the community where they do (or don’t) match your experiences! You can find more information on where to do this at the end.


The power of a semantic layer on top of a mature data modeling framework

As a longtime dbt Community member, I knew I had to get involved when I first saw the dbt Semantic Layer in the now infamous dbt should know about metrics Github Issue. It gave me a vision of a world where metrics and business logic were unified across an entire organization; a world where the data team was no longer bound to a single consuming experience and could enable their stakeholders in dozens of different ways. To me, it felt like the opportunity to contribute to the next step of what dbt could become.

In past roles, I’ve been referred to as the dbt zealot and I’ll gladly own that title! It’s not a surprise - dbt was built to serve data practitioners expand the power of our work with software engineering principles. It gave us flexibility and power to serve our organizations. But I always wondered if there were more folks who could directly benefit from interacting with dbt.

The Semantic Layer expands the reach of dbt by coupling dbt’s mature data modeling framework with semantic definitions. The result is a first of its kind data experience that serves both the data practitioners writing your analytics code and stakeholders who depend on it. Metrics are the first step towards this vision, allowing users to version control and centrally define their key business metrics in a single repo while also serving them to the entire business.

However, this is still a relatively new part of the dbt toolbox and you probably have a lot of questions on how exactly you can do that. This blog contains our early best practice recommendations for metrics in two key areas:

  • Design: What logic goes into metrics and how to use calculations, filters, dimensions, etc.
  • Structure: Where these metrics will live in your dbt project and how to compose the files that contain your metrics

We developed these recommendations by combining the overall philosophy of dbt, with our hands-on learning gathered during the beta period and internal testing.

· 11 min read

When you were in grade school, did you ever play the “Telephone Game”? The first person would whisper a word to the second person, who would then whisper a word to the third person, and so on and so on. At the end of the line, the final person would loudly announce the word that they heard, and alas! It would have morphed into a new word completely incomprehensible from the original word. That’s how life feels without an analytics engineer on your team.

So let’s say that you have a business question, you have the raw data in your data warehouse, and you’ve got dbt up and running. You’re in the perfect position to get this curated dataset completed quickly! Or are you?

· 10 min read

Why do people cherry pick into upper branches?

The simplest branching strategy for making code changes to your dbt project repository is to have a single main branch with your production-level code. To update the main branch, a developer will:

  1. Create a new feature branch directly from the main branch
  2. Make changes on said feature branch
  3. Test locally
  4. When ready, open a pull request to merge their changes back into the main branch

Basic git workflow

If you are just getting started in dbt and deciding which branching strategy to use, this approach–often referred to as “continuous deployment” or “direct promotion”–is the way to go. It provides many benefits including:

  • Fast promotion process to get new changes into production
  • Simple branching strategy to manage

The main risk, however, is that your main branch can become susceptible to bugs that slip through the pull request approval process. In order to have more intensive testing and QA before merging code changes into production, some organizations may decide to create one or more branches between the feature branches and main.

· 10 min read

If you’ve ever heard of Marie Kondo, you’ll know she has an incredibly soothing and meditative method to tidying up physical spaces. Her KonMari Method is about categorizing, discarding unnecessary items, and building a sustainable system for keeping stuff.

As an analytics engineer at your company, doesn’t that last sentence describe your job perfectly?! I like to think of the practice of analytics engineering as applying the KonMari Method to data modeling. Our goal as Analytics Engineers is not only to organize and clean up data, but to design a sustainable and scalable transformation project that is easy to navigate, grow, and consume by downstream customers.

Let’s talk about how to apply the KonMari Method to a new migration project. Perhaps you’ve been tasked with unpacking the kitchen in your new house; AKA, you’re the engineer hired to move your legacy SQL queries into dbt and get everything working smoothly. That might mean you’re grabbing a query that is 1500 lines of SQL and reworking it into modular pieces. When you’re finished, you have a performant, scalable, easy-to-navigate data flow.

· 14 min read

Analyzing financial data is rarely ever “fun.” In particular, generating and analyzing financial statement data can be extremely difficult and leaves little room for error. If you've ever had the misfortune of having to generate financial reports for multiple systems, then you will understand how incredibly frustrating it is to reinvent the wheel each time.

This process can include a number of variations, but usually involves spending hours, days, or weeks working with Finance to:

  • Understand what needs to go into the reports
  • Model said reports
  • Validate said reports
  • Make adjustments within your model
  • Question your existence
  • Validate said reports again

You can imagine how extremely time consuming this process can be. Thankfully, you can leverage core accounting principles and other tools to more easily and effectively generate actionable financial reports. This way, you can spend more time diving into deeper financial analyses.

· 5 min read

Semantic layer, Python model support, the new dbt Cloud UI and IDE… there’s a lot our product team is excited to share with you at Coalesce in a few weeks.

But how these things fit together—because of where dbt Labs is headed—is what I’m most excited to discuss.

You’ll hear more in Tristan’s keynote, but this feels like a good time to remind you that Coalesce isn’t just for answering tough questions… it’s for surfacing them. For sharing challenges we’ve felt in silos, finding the people you want to solve them with, and spending the rest of the year chipping away at them. As Tristan says in his latest blog, that’s how this industry moves forward.

REGISTER NOW

· 7 min read

Editors note - this post assumes working knowledge of dbt Package development. For an introduction to dbt Packages check out So You Want to Build a dbt Package.

It’s important to be able to test any dbt Project, but it’s even more important to make sure you have robust testing if you are developing a dbt Package.

I love dbt Packages, because it makes it easy to extend dbt’s functionality and create reusable analytics resources. Even better, we can find and share dbt Packages which others developed, finding great packages in dbt hub. However, it is a bit difficult to develop complicated dbt macros, because dbt on top of Jinja2 is lacking some of the functionality you’d expect for software development - like unit testing.

In this article, I would like to share options for unit testing your dbt Package - first through discussing the commonly used pattern of integration testing and then by showing how we can implement unit tests as part of our testing arsenal.

· 12 min read

Those who have been building data warehouses for a long time have undoubtedly encountered the challenge of building surrogate keys on their data models. Having a column that uniquely represents each entity helps ensure your data model is complete, does not contain duplicates, and able to join across different data models in your warehouse.

Sometimes, we are lucky enough to have data sources with these keys built right in — Shopify data synced via their API, for example, has easy-to-use keys on all the tables written to your warehouse. If this is not the case, or if you build a data model with a compound key (aka the data is unique across multiple dimensions), you will have to rely on some strategy for creating and maintaining these keys yourself. How can you do this with dbt? Let’s dive in.

· 15 min read

The larger a data ecosystem gets, the more its users and stakeholders expect consistency. As the ratio of data models to team members (to say nothing of stakeholders to team members) skyrockets, an agreed-upon modeling pattern often acts as scaffolding around that growth.

The biggest tool in the toolbox today, dimensional modeling, offers enough consistency to make it the dominant approach in the space, but what might be possible if we shut that toolbox, took a break from our workbench, and instead strolled over to our bookshelf?

In other words, what if we told a story?

· 15 min read

When running a job that has over 1,700 models, how do you know what a “good” runtime is? If the total process takes 3 hours, is that fantastic or terrible? While there are many possible answers depending on dataset size, complexity of modeling, and historical run times, the crux of the matter is normally “did you hit your SLAs”? However, in the cloud computing world where bills are based on usage, the question is really “did you hit your SLAs and stay within budget”?

Here at dbt Labs, we used the Model Timing tab in our internal analytics dbt project to help us identify inefficiencies in our incremental dbt Cloud job that eventually led to major financial savings, and a path forward for periodic improvement checks.

· 13 min read

At dbt Labs, we have best practices we like to follow for the development of dbt projects. One of them, for example, is that all models should have at least unique and not_null tests on their primary key. But how can we enforce rules like this?

That question becomes difficult to answer in large dbt projects. Developers might not follow the same conventions. They might not be aware of past decisions, and reviewing pull requests in git can become more complex. When dbt projects have hundreds of models, it's hard to know which models do not have any tests defined and aren't enforcing your conventions.