The list of connectors is quite impressive which is very important for data integration. But I could not find what kind of data transformation operators are available for building an ETL or data migration workflow.
I'm an engineer at Stitch. Our approach to transformation is to do just enough to move data from one system to another without losing precision or fidelity. So, we transform datatypes and structures into more appropriate forms for the target system, but we don't have any transformation operators like aggregation or windowing.
We have found that this approach works well for our users, who prefer to get the rawest possible data, and the systems we target like Redshift that are themselves powerful transformation engines. This gives the user unlimited flexibility for defining transformations, and a full audit trail for understanding how their data has changed.
We are always evolving, though, so if there's a use case that you think requires this approach, I would be eager to hear more about it.
Do you have any sort of SDK for adding integrations that you do not support?
While this looks super useful if you support all of the integrations someone needs, it seems like the moment that's not the case someone needs to maintain a complete ETL pipeline for those data sources you don't support, and their load is only reduced by the fact that they have to maintain fewer data sources.
I have no idea what you're talking about. Scanning your docs, I'm no more illuminated.
I've done a lot of ETL, mostly for healthcare.
Yes, engineers should be doing ETL work. Any "workflow engine" that promises patch cord or visual programming is hooey. At the end of the day, someone somewhere is gonna be writing some code. And its not the "business analyst" or "subject area expert". No, its a dev. And all that clever framework stuff is just an angry 800lb gorilla sitting between her and her work.
ETL is just fancy talk for data processing. Input, processing, output. Copy a string from a source, maybe mangle it a bit, paste that string somewhere else. Extra credit for type awareness, eg "oh! that string's a date!". Trophies for logging, alerts, and services which heal themselves.
>nobody enjoys writing and maintaining data pipelines or ETL. It’s the industry’s ultimate hot potato. It really shouldn’t come as a surprise then that ETL engineering roles are the archetypal breeding ground of mediocrity.
Bullshit. I enjoy it.
Yes, I'm a mediocre programmer, but I love building ETL. I'm also great at dinner parties.
this might be valuable technology, but as far as i can from reading the OP and from a look through the repo, this is not ETL--despite the fact that "ETL" is in blog post's title, despite the fact that the technology is described there as "a fully managed ETL service", and despite the fact that the term is used over a dozen times in the short blog post. As everyone here knows, ETL is "extract transform load"; this technology is directed to a portion of the first of those three, "extact", and indeed" only a portion because the sole (?) focus seems to be on providing various connectors for data access from various sources, but Stitch doesn't for instance, provide a query language API so that you can write a single query that works across multiple data sour es.
granted, i think providing a library of connectors for many diverse, data sources--eg, SalesForce CRM, Google Analytics, MySQL--is valuable, but it's not ETL, it's not even "E", just part of it. What's more the functionality Stitch does have is, in my experience, far from the most difficult or time-consuming component to build, instead, "Transform" is.
It's accurate to say that we're more focused on the extraction and loading parts of ETL. In our experience, almost all useful data analysis requires data to be transformed at multiple stages - for example, once to cleanse and normalize the data, and again to aggregate it. Our goal is to leave the data in the rawest form possible without losing accuracy or precision, so that further transformation can occur, likely using SQL. We recognize this isn't perfect for every use case, but our customers love it for getting rapid access to data that would otherwise be locked away in SaaS applications and transactional databases.
9 comments
[ 5.4 ms ] story [ 22.0 ms ] threadWe have found that this approach works well for our users, who prefer to get the rawest possible data, and the systems we target like Redshift that are themselves powerful transformation engines. This gives the user unlimited flexibility for defining transformations, and a full audit trail for understanding how their data has changed.
We are always evolving, though, so if there's a use case that you think requires this approach, I would be eager to hear more about it.
While this looks super useful if you support all of the integrations someone needs, it seems like the moment that's not the case someone needs to maintain a complete ETL pipeline for those data sources you don't support, and their load is only reduced by the fact that they have to maintain fewer data sources.
Additionally, we'll be releasing a Java client library any day now, with other languages and platforms to follow.
I've done a lot of ETL, mostly for healthcare.
Yes, engineers should be doing ETL work. Any "workflow engine" that promises patch cord or visual programming is hooey. At the end of the day, someone somewhere is gonna be writing some code. And its not the "business analyst" or "subject area expert". No, its a dev. And all that clever framework stuff is just an angry 800lb gorilla sitting between her and her work.
ETL is just fancy talk for data processing. Input, processing, output. Copy a string from a source, maybe mangle it a bit, paste that string somewhere else. Extra credit for type awareness, eg "oh! that string's a date!". Trophies for logging, alerts, and services which heal themselves.
Bullshit. I enjoy it.
Yes, I'm a mediocre programmer, but I love building ETL. I'm also great at dinner parties.
granted, i think providing a library of connectors for many diverse, data sources--eg, SalesForce CRM, Google Analytics, MySQL--is valuable, but it's not ETL, it's not even "E", just part of it. What's more the functionality Stitch does have is, in my experience, far from the most difficult or time-consuming component to build, instead, "Transform" is.