Cubegeek

Eye Candy

 

 

 

I must confess that there’s a lot of cool stuff out there. If I were a PhD or genuine science-type computer scientist I would be super stoked about the revolution. But since I am a humble practitioner, it’s a bit of hell, especially because I can’t tell which of these hot technologies (or are they cool?) are going to be significant in the market. You know, so I can bank money on it. So here’s a top five, among all of them that I find the most interesting and am going to put the most time and effort into gaining a comprehensive understanding.

DuckDB
This package continues to delight and satisfy. It’s changing my ideas, which you already know, about “Try this at home”. because its extensions are so wonderful. With or without MotherDuck, this thing rocks. Now it turns out that it has a UI built in and it scans DeltaLake and it has something called DuckPGQ.

SlateDB
Just found out about this this morning. Brilliant idea which could obviate S3 Tables. I have particularly liked GBQ because if your data just sits for a long time, it’s cheap to operate. I don’t know why but that’s one of the things I love about it. I’ve been using GBQ to catalog my (small) stuff and paying next to nothing. But if SlateDB does what it claims, then I can keep my stuff in R2 (just recently discovered, and almost as good as B2 or S3) then we get even closer to free.

DeltaLake & Unity Catalog
I just think this is going to be the winner. Databricks is a thought leader and the more they push open standards, the more they will win as well. Still, CockroachDB is more approachable.

MCP
The thing that made the most sense is that data pipelines are going to get more complicated as people kludge AI + ML + DW + Streams. It’s all going to get ugly, but MCP is going to be the next API of the future, even if people are just asking it stupid things. So I think it’s going to be the thing to master in the near term, even though it will be impractical for the next six months.

Now here are the guys I’m reading because I’m just scared not to.

• Daniel Beach at 

  Daniel Beach

• Chris at 

  Materialized View

• Nate 

  Nate’s Substack

   (but what’s up with that hat, dude?)

• Asianometry

• AI Explained

Restart (Again)

While most of my work is proprietary, I have come into several good reasons for thinking out loud about skills and practice. And since I own this domain, I am doing so as part and parcel of establishing some training 'influencing' in this new and exciting (and confusing) time in all things data. So I will be talking about MCP and data cataloging and small (monolithic DuckDB) data. In the short term you can alternate between here at Tessellations which is going to be some duplication back and forth, because data and publication is cheap.