castorp

I was talking about a backup, not a dump (snapshot).

https://blog.dbi-services.com/what-is-a-database-backup-back...

> I know there are some ways that reduce the downtime, like hard linking the data between the clusters instead of version

You need to have a proper backup of the database anyway, so I don't really see the risk in using the --link mode (which makes the downtime pretty much independent of the size of the actual data).

Well, there was the case where the the NSA intercepted hardware shipments (I think it was Cisco HW) to install their backdoors.

I think you will always find something that the other product doesn't have (or doesn't have in the same way)

Postgres e.g. has the infrastructure for extendable data types and index types and the infrastructure for extensions - none of that can be found in MySQL and that infrastructure needs source code as well.

I don't think the nearly 3-fold difference can't be attributed to "features", but to other things. Maybe the projects are structured differently or include different additional tools in their code (e.g. the input files to generate the SQL parser, rather than the generated parser's C code).

> Zheap does lead to lots of tricky scenarios. If for any reason you need to access the old copy of the row, you have to fetch it from the separate file. If the transaction that performed that update is rolled back, you need to replace the new version of the row with the old version of the row. This sounds straightforward, but gets really tricky really fast.

This is pretty much what Oracle is doing.

The SQL standard also defined an operator for that: IS DISTINCT FROM

There are some areas where Oracle indeed offers more features (mainly in the administration/monitoring area) compared to Postgres, but PL/SQL vs. PL/pgSQL as a procedural language PL/pgSQL is not one of them. Both languages are pretty much on par

That's plain wrong. Ora is short for Oracle

> 4. Parallel Query Execution

Available since Postgres 9.6 and getting better with every release.

> How do people keep track on all the features that are added in RDMS?

By reading the release notes

> One important point is that Uber wrote this running on Postgres 9.2. The latest Postgres release is now 12.4

Just a minor nitpick: Postgres changed the version numbering scheme with version 10. So 9.4 was a major release, but 12.4 marks a minor (bugfix) release.

If you want to name minor releases it would be e.g. 9.4.26 vs. 12.4. If you want to compare major versions it would be 9.4 vs. 12

The Postgres team decided against implementing MERGE when they designed the ON CONFLICT feature (it actually started as a MERGE).

It's been a while, but as far as I remember, the main reason was that the behaviour of MERGE mandated by the SQL standard, wasn't clear enough and doesn't deal with race conditions when multiple transactions try the same insert (or update) (not so much the syntax itself).

> At the end of the day, you still have to go to disk on writes, and this must be serialized against reads for basic consistency reasons.

No, absolutely not.

That's why modern databases use a thing called multi version concurrency control. You can run (multiple) queries on the same table that is updated by multiple transactions at the same time without one blocking the others (assuming the write transactions don't block each other). Of course they are fighting for I/O, but there is no need so serialize anything.

Mixing OLTP and OLAP becomes increasingly "normal" theses days as the capabilities of the database products and the hardware improve. With modern high-end hardware (hundreds of CPUs, a lot of SSDs, large RAM) this actually scales quite nicely .

> Boeing will build a new version of its F-15, an older fighter jet developed in the 1960s

Am I the only one that immediately thinks about the 737max disaster where they tried the same thing?

Postgres does not support read uncommitted

There is a built-in supported set of features for high availability. What exactly are you missing?

I'll test it as well:

- stupid dark mode: makes me cranky, gives me a headache - light mode: relaxed reading, makes me feel comfortable. - different results ever? No.

the above is from the undisputed, worldwide export on what I feel.

I absolutely hate dark mode.

I even had to install a Firefox plugin to disable dark colors, to make website that use it readable again.

The Postgres team has been using SQLSmith for that purpose for a while if I'm not mistaken.

What's the difference between a query hint and a planner hint?

> automatic planner adjustment when true cost was off

More often than not, this "automatic adjustment" makes things worse.

> active-active clustering with RAC

But it's still shared storage, which doesn't really make it a true cluster in my opinion

> out of the box backup tooling

While this is true, there are several (free) backup tools that are easy to install and use and can rival rman's features.

> Very few databases support transactional DDL

Postgres, SQL Server, Firebird, DB2, SQLite, Ingres, Informix, Teradata, Vertica all support transactional DDL

> MySQL and PG are not truly consistent per default, they don't fsync every writes.

Postgres most certainly does fsync by default.

It's tru, you can disable it, but there is a big warning about "may corrupt your database" in the config file.

How does that look like with a multi-row insert?

And when you need to insert more than one row, you repeat the column name over and over again for each row - not a good idea either.

Are there any European hosted (and owned by a European company) alternatives to GitHub or GitLab?

But a lot more people find loud engines simply horrible and inconsiderate.

I was referring to the native execution using LLVM. Which should be independent of the underlying storage strategy (column-store, row-store)

Can it do load balancing and failover?

Well, this simply moves the problem of "commenting out" one column from the end of the list to the start of the list. I find myself doing both equally frequent.