“Nothing about this means tools like Fivetran+dbt are suddenly open source. The idea is that they operate on top of infrastructure that is less restrictive than the old warehouse model - there’s so much to unpack tho in terms of current technological developments and what future data platform will look like.”
“Because they typically price on monthly deltas volumes there’s kind of a middle tier where it makes sense as part of your tech stack. Too low and it’s too expensive, and if your data volumes are massive, again, too expensive. But if you’re in that sweet spot, it may be worth paying a vendor than paying an engineer to perform those tasks.”
“All that said, my Fivetran and dbt Cloud bill is much higher than my Snowflake bill so I'm not worried about the compute layer like they seem to think companies are.”
“OpenFlow for all things on-prem SQL Server which is 90% of what we are replicating. We will likely leave the other connectors on Fivetran until OpenFlow has connectors or we find another vendor that would make it easy to switch to.”
“My suggestion would be dlt. It's open source, but dlthub, the company behind it, is building something on top of it, not just a dlt SasS, so I am pretty confident it will stay open.”
“We also use DBT and have been using the transformations inside of fivetran so they want to entertain the idea of finding something similar (probably not possible I told them)”
“We offer hundreds of connectors, and the AI makes it dead simple to start immediately building out analyses or dashboards.A bit about what’s going on under the hood:- We use Fivetran to connects to data sources (we tested at a number of different ETL solutions and Fivetran really stood out)”
“Are you personally planning to fork SQLMesh and spend a substantial chunk of your time maintaining it, if Fivetran strip the original for parts? Perhaps you will fund this via a hosted version - will you be picking up the phone to pitch it to companies yourself, or will you use your own money to pay someone to do that?”
“I think they're going to move into the compute layer and offer the ability to run your dbt transformations using Fivetran compute (likely powered by DuckDB) at the same price or cheaper than you'd get with Snowflake/Databricks/BQ.”
“Fivetran is great for fully managed, low-maintenance pipelines; Airbyte offers flexibility and open-source customization; DLT is promising for more modular, developer-centric workflows.”
“With 80-90% customer overlap and pricing already strained (users report 4-8x cost increases under Fivetran's recent pricing changes), expect layoffs and further price optimization as the merged company eliminates redundancies.”
“(2) Fivetran is over-priced - Although I do think that Fivetran is an industry-leading product, it is also highly over-priced. It seems inevitable that this commercial strategy will eventually extend to SQLMesh and dbt.”
“Just like Plaid's approach became the standard to code against in accessing financial services, Fivetran has become the default in getting data out of tools you already pay for directly into a controlled space.If they don't muck that up, they've got an embed in every large integration.”
“Fivetran is making money hand over fist on ingestion. Together, dbt adoption can drive revenue growth without dbt itself being monetized as orgs default more to using Fivetran to load data because it "just works" with dbt and expand to new sources because they're able to actually use more data efficiently”
“It feels like betting on the wrong horse, fivetran is pricing itself out of a market. I've been on two projects where we setup custom ingestion pipelines to save money on fivetran, and we expect to make the salary spend back within a year.”
“Fivetran is convenient but absurdly expensive. Now dbt core development is just another marketing cost for Fivetran.I'd really love to see everyone move on from the "modern data stack".”
“Fivetran acquired Census (reverse-etl) & Tobiko (dbt alternative).I wonder who's next to really consolidate their platform play and compete with the old legacy MDM provider like Informatica.”
“It makes sense with Fivetran and dbt being so complimentary, ingestion (Fivetran) vs transformation (DBT).But I still feel like I'm missing, in DBT, capabilities for DB DDL/DML deployment (which I've done from Liquibase in DBT) for a fully CICD modern data stack.”
“Fivetran will give you an estimate of MAR during your 14 day free trial. I suspect, based on your comment re: “basic employee data” you’ll be below 500K MAR, and will qualify for the free tier.”
“I've never tried fivetran, open sweat solutions are what I use here, apache hop Airbyte, pentaho and python, no problem extracting several times a day.”
“fivetran's pricing can be tricky. free tier is limited, and with consistent usage, fees might creep up. consider sticking with python or explore other affordable etl tools.”
“Currently using Python to extract data from our HR tool Personio through REST API. Now I saw that fivetran is offering a connector, so I am thinking about switching to easen the extraction process.”
“You’re right that History Mode aims to model SCD2, but in practice it’s closer to a lossy implementation. It tracks versioned rows and soft deletes, but lacks true bi-temporal semantics, no valid\_from/valid\_to, just \_fivetran\_synced snapshots.”