Pick a replacement by who will own the pipelines. Engineers comfortable in Python or the command line can run open source tools as code, while smaller teams without a data engineer are better served by a managed or no-code service that looks after the connectors for them. Seven options follow, in no particular order.
Connectors are the whole product in this category. Portable's homepage leads with a very large catalogue of supported sources, which is why teams with unusual, long-tail apps often end up there. Leaving it means checking, source by source, that the apps you rely on are covered somewhere else, and deciding how much of the plumbing you're happy to own yourself. The tools below cover that whole range, from a library you write code against to services you configure in a browser.
Seven ELT and pipeline options in alphabetical order
In no particular order of preference. Each line sums up the vendor's own description of its product.
dlt
Best for: Python developers who'd rather write pipelines as code; the library is open source, and the vendor's hosted platform deploys, monitors and scales them.
Fivetran
Best for: Teams that want data from many apps centralised for analysis with minimal effort; the vendor now also pitches connecting AI to that data, and names a free plan.
Hevo Data
Best for: Analysts and small data teams who want to load data into a warehouse, transform it and send it on to other databases, per the vendor.
Keboola
Best for: Companies whose leadership wants AI but whose data isn't ready; the vendor pitches a governed data foundation aimed at finance chiefs.
Meltano
Best for: Engineers who want vendor-neutral, open source pipelines they own and run in their own cloud as code, according to the vendor.
RudderStack
Best for: Groups focused on customer data, from collection through to governance, on a platform the vendor calls warehouse-native; a free plan is named.
Skyvia
Best for: Non-developers who need cloud data integration without code, plus cloud-to-cloud backup and SQL access to their data, as the vendor lists it.
Where each one sits
| Tool | Approach | Good for |
|---|---|---|
| dlt | Open source Python library | Python developers building pipelines in code |
| Fivetran | Managed data movement | Centralising app data with little engineering effort |
| Hevo Data | Managed ELT platform | Loading and transforming data without heavy engineering |
| Keboola | Governed data foundation | Finance-led companies preparing data for AI |
| Meltano | Open source data integration | Engineers who want pipelines in code they control |
| RudderStack | Customer data platform | Collecting and governing customer data in the warehouse |
| Skyvia | No-code cloud data platform | No-code integration, backup and SQL access |
The question that decides it: who fixes a broken connector?
Every pipeline tool looks fine on the day you connect your first source. The real test comes a few months later, when a software vendor changes its API and a sync quietly stops. With a managed service, the vendor's team usually maintains the connector and you wait for a fix. With an open source library, you or your engineers patch it, which is faster if you have the skills and painful if you don't.
So look at three things for each tool. First, which of your sources are supported today, by name, and whether long-tail apps are covered or would need a custom connector. Second, how syncs are scheduled, and whether you can see when one fails before someone notices a broken dashboard. Third, who maintains each connector afterwards. Run the same real task through each shortlisted tool, such as moving one awkward source into your warehouse, before you decide.
One familiar name has shifted. Stitch's own homepage now points visitors towards Qlik Talend Cloud, so it isn't listed here as a separate option.
Code-first, managed or no-code options
Code-first. dlt and Meltano are open source and built for engineers who want pipelines in version control, running in their own environment. You get control and you take on the upkeep.
Managed services. Fivetran, Hevo Data and Keboola handle the plumbing for you, with different emphases. One focuses on centralising data with very little effort, one on loading and transforming it for analysts, and one on governance for companies getting their data ready for AI.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Customer data and no-code. RudderStack centres on customer data, from collection to governance, which suits teams whose main sources are their own website and app events. Skyvia is the pick for non-developers, with integration, backup and SQL access handled without code.
My advice before you switch pipelines
Don't migrate everything in one weekend. Move one source at a time, run old and new syncs side by side for a short period, and compare row counts and a few key reports before you switch the old one off. It's tedious, but it's the only way to be sure nothing has gone missing on the way.
Be realistic about engineering time, too. A code-first tool can feel like the flexible option, but someone has to own it when it breaks at an inconvenient moment. If that person doesn't exist yet, a managed service is the safer start. Good pipelines are only worth having if the reporting on top is used; my notes on what makes business intelligence work and on using AI in business intelligence cover that side.
What to ask on every vendor demo
- Which of our sources do you support today, by name, and who built each connector?
- What happens when a source changes its API: who notices, and who fixes it?
- How are syncs scheduled, and how are we told when one fails?
- Can we transform data inside the tool, or does that happen in the warehouse afterwards?
- If we leave, can we take our pipeline definitions with us?
Do you make data pipeline or ELT software? Tool makers can ask to be considered for a featured place on this list.
Want your tool on this list? Featured spots and reviews here are paid partnerships. Here's how to partner with me.
Related: the python page.
Frequently asked questions
What is an ELT tool?
An ELT tool extracts data from your apps, loads it into a warehouse and lets you transform it there. It saves you writing and maintaining your own scripts for each source.
What's the difference between ETL and ELT?
ETL transforms data before it reaches the warehouse; ELT loads raw data first and transforms it inside the warehouse. Most modern cloud tools favour ELT.
Should I use an open source pipeline tool?
It's a good fit if you have engineers who are happy to run and patch it. If nobody on the team owns pipelines, a managed service will usually be less stressful.
How do I check a tool supports my data sources?
Ask for each source by name, and ask who maintains the connector. A connector listed on a page isn't the same as one that's kept up to date.
Can a small business use a data warehouse?
Yes. Many small teams start with a cloud warehouse and a handful of sources, and add more as reporting needs grow.
How long does it take to switch pipeline tools?
It depends on the number of sources. Moving one at a time and running old and new syncs in parallel for a while is slower but much safer.
