Why Materialize: Recommended Reading List
A collection of articles that convinced me about the necessity and long-term inevitability of a technology like Materialize.
Image credit: Factory (circa 1920) Preston Dickinson (American, 1891-1930)
This is a growing collection of articles, podcasts, and other media that convinced me that:
- Technology that incrementally maintains materialized views of data is difficult, but necessary and inevitable
- Materialize is the best shot at making that available to a wide array of developers.
These are probably most relevant to people with operational experience in the data space. You don’t need to be technical but it’s useful if you’ve rolled up your sleeves and dug into some of the complicated bits of working with data at scale. They don’t need to be read in a particular order.
Listen to Materialize cofounder Arjun Narayan’s podcast on the Data Stack Show: A New Paradigm in Stream Processing
Watch our other cofounder and chief scientist Frank McSherry’s DataCouncil talk on Materialize to get a feel for how Materialize works.
Read Jon Gjengset’s “simple english” explanation of batch vs incremental https://jon.thesquareplanet.com/noria-in-simpler-terms.pdf - This is just a really great non-technical way of explaining the traditional database vs incrementally-maintained db with a fun metaphor. Jon created his own approach to incremental view maintenance as an MIT thesis, but has since moved on to other things.
Read Nikita Tonsky’s “Web After Tomorrow” - Nikita is thinking about things from an application developer’s perspective, looking for a “reactive database” but it is eye-opening. Be warned that we are still a ways away from developing a product that is good enough to just quickly slot in and solve this problem.
Read Data Denormalization is Broken - This is a good practical explainer of why something like Materialize should exist.
Read A Data Pipeline is a Materialized View - If the one above is for application development, this post lays out the case for Materialized Views for data engineering.
Read Tristan Handy’s post/rant about the limits of incrementally updated tables in data warehouses. - This is more of an answer to “Why NOT just work with the paradigms we have."
Once you’re finished, if you’re excited about Materialize, check out our jobs page, if you’re interested in building on-ramps to Materialize by contributing both writing and code, read about our developer experience team and reach out if it interests you!