SQL

It's by previous guest and creator of the Narwhals Library, Marco Gorelli.

Uh, it's titled The Polars versus SQL Differences Nobody Is Talking About.

Some of these crucial differences that he is highlighting here, I'll read from the second paragraph.

It's titled Mental Models.

[laughs] This allows for column independence.

Row order is defined, but as we will also see later, Polars doesn't always promise to preserve it.

SQL is a programming language used to interact with databases which follow the relational model.

In the relational model, a table is an unordered bag of rows, each of which is atomic, meaning that it can't be split up.

After that, we will head over to Kubernetes, discussing the release of GKE 1.37 and a container-eyed privilege escalation vulnerability.

Then we will cover image updates in Managed Service for Apache Spark, Cloud SQL's regional endpoints going GA, and big updates in networking, identity, and cloud storage.

We have a lot of ground to cover, so let's get right into it.

First up, let's talk about BigQuery.

So make sure to connect using a standard non-root user like Dataproc and run commands with sudo.

Now let's dive into some database and serverless updates.

First up, regional endpoints are now generally available for Cloud SQL, MySQL, PostgreSQL, and SQL Server Admin APIs.

Instead of hitting a single global endpoint, You can now interact with your Cloud SQL instances using regionalized URLs like SQL admin region representative googleapis.com.

That's why in memory caches became so popular.

Oh, what do you mean this SQL query is so slow? and I can just cache it? Wow, that sounds amazing.

So instead of taking time to understand why something is slow, we took the shortcut of executing it once or maybe per process and then cache the results somewhere so someone else can consume the results.

Inevitably with caching comes stale results, comes inconsistent results, comes out of date management of the cache.

actually there is a great video by Dave from he used to be a Microsoft developer he's retired now but he went through how they used to optimize Windows was it 95 or 98 I can't remember that the executable from Windows at night they take the executable and run it and see how the functions are called and then put the functions that are close to each other in the same four kilobyte pages and that's because that's how cpu execute pages right ah thank you train it's okay you don't have to honk so yeah sometimes the train is just like the conductor like to honk for no reason i don't know why this just feels like it there's there's nothing here in this area but anyway so By optimizing that, every memory access gets you one beautiful page and every subsequent calls are almost always in this area.

So you see how precious these calls get.

So back to our life, how we do things is when we execute... a REST call or a GraphQL or any endpoint or a SQL statement, we just assume that, you know what, I'm not going to do that again.

That's slow.

And what if it fails?

Let's say the script throws in SQL syntax error.

The AI reads the error log, realizes the company uses some non-standard database structure, rewrites its own script, and tries again.

It iterates through the errors in seconds, navigating the proprietary environment until it successfully completes the objective.

And in fairness, you know, we're technical people, but I think about all of the care and feeding that I've given Claude code over the last, you know, year.

And I'm looking at these outputs and I'm looking at this div and I'm looking at this table and I'm looking at this SQL output.

And, how often am I actually involving myself in that decision-making? Like I know enough to be dangerous, but I'm not reading every line of the code.

I'm looking for the output.

There's plenty of other stuff that we can be using AI for right now in your business to help achieve bigger outputs and levers for you.

Yeah, yeah, yeah, definitely.

But there's also security risk, like, you know, that there is I've joked about the AI plumbing thing in the past, like, I would say that, you know, SQL has come like Snowflake has come a long way.

And we used to do all of this stuff where we would be extracting data from a Snowflake database, we'd be pulling it out, we'd be transforming it, and then we'd be shoving it back in.

So the software is telling you the correct order of operations if you know how to hear it.

Now, we have data and tables connected by keys at this point, right? So how do we get anything out? well you ask a question and the formal name for that question is a query the most common language for writing one is sql or sql and nobody went and when you're talking about whether it's called sql whether it's called sql no one's going to win that argument so just call it what you want to call it Now I'm not going to teach you SQL in three minutes, but I want you to recognize the shape of it.

So when you see a query in a support ticket or an error log, you can read it well enough to know what it was trying to do.

almost and almost every query is built from six pieces you have select which says which fields you want back so like give me the point name and value there's from which says which table to pull from so from the points table example Where filters the row, where, W-H-E-R-E, only where the unit is age U3, right? So this piece does the most work, and it is where performance problems tend to show up.

But there isn't any equivalent in data.

And what we've seen in the past, like in the data world pre-AI, so like two years ago, before two years ago, something like this, everyone was writing SQL every day.

Like when you take like someone in the data team, you were writing like a few SQL query from like five to 50 to 100 a day, something like this, editing SQL queries all day long.

And at the beginning we were like, yeah, this is what the world is.

People are writing SQL query.

We're going to help people write SQL query better, faster, without errors, with autocomplete and so on.

But like world shifted super fast when Cloud Code arrived.

And what we've seen is people don't even care looking at the data anymore.

Okay, well, let me decide what database we should pick.

Obviously, SQL, like, no SQL is, like, not great.

But I had to make all these, like, really big decisions.

And Warp's backend is still running on, like, the decisions that I made.