nodejs-bigquery VS corp

If you've ever wondered what it's like to manage a BigQuery instance at Reddit scale, know that it's exactly like smaller systems just with much, much bigger numbers in the logs. Database management fundamentals are eerily similar regardless of scale or platform; BigQuery handles just about anything we throw at it, and we do indeed throw it the whole book. Our BigQuery platform is more than 100 petabytes of data that supports data science, machine learning, and analytics workloads that drive experiments, analytics, advertising, revenue, safety, and more. As Reddit grew, so did the workload velocity and complexity within BigQuery and thus the need for more elegant and fine-tuned workload management.

Now I know I've got some data I could use, I now need to find a platform that I can use to analyse the data coming from the Forem API. I did consider some other pieces of software, such as Google BigQuery (with looker studio) and ElasticSearch (with Kibana), I ultimately went with OpenSearch which is essentially a forked version of ElasticSearch maintained by AWS. The main reasons are that I could host it locally for free (unlike BigQuery). I do have some prior experience with both elastic (back when it was called ELK) and OpenSearch, but my work with OpenSearch was far more recent, so I decided to go with that.

First, in one of our recent projects, we helped our client to run the cloud-based infrastructure of their entirely automated, real-time SEO platform. The solution rested in the safe familiarity of Google’s popular cloud-based data centres (i.e. Google Cloud Platform), whilst also making use of BigQuery — a serverless, multi-cloud data warehouse.

Bigquery as our data warehouse

This post in github thinks it may be an error in bigquery's backend.

A data warehouse is a specialized database that's purpose built for gathering and analyzing data. Unlike general-purpose databases like MySQL or PostgreSQL, which are designed to meet the real-time performance and transactional needs of applications, a data warehouse is designed to collect and process the data produced by those applications, collectively and over time, to help you gain insight from it. Examples of data-warehouse products include Snowflake, Google BigQuery, Azure Synapse Analytics, and Amazon Redshift — all of which, incidentally, are easily managed with Pulumi.

Google BigQuery

Here's one that covers some of your points and that I like in general: https://github.com/dbt-labs/corp/blob/main/dbt_style_guide.md Except instead of prefixing my table names with the processing stage, I keep them in schemas by processing stage (source, staging, analytics). So, I can tell my analysts to look into the analytics schema for all the final tables, and they won't be bothered by intermediate models. The table names also have a precise structure that corresponds to our specific subject.

I find the content more or less ripped from of dbt's own styleguide

For anyone wondering this is the DBT style guide I am referencing from.

While the tool choice is obvious, how to use dbt is going to be a more controversial. There's a load of great resources on dbt best practices, but as you can see from my Slack questions, there's enough ambiguity to tie you up.

Just a slight critique, but I noticed some of the dbt models are a bit hard to read. Especially your dim_users SCD2 model, which uses lots of nested subqueries and multiple columns on the same line. You may want to refer to this style guide from dbt Labs. I find CTEs are a lot easier to parse and read.

I really like thisthis SQL STYLE GUIDE, and if you use dbt, the dbt style guide.

I like this one very much from dbt very much.

I recommend you write one. You can take inspiration from dbt's one or Gitlab

I've seen fishtown analytics coding conventions recommend a lot around here, but there are a few things about their recommendations of CTE use that confuse me.