glogla

Snowflake and ClickHouse are very different in their focus.

Snowflake is focused on enterprise customers. It has a lot of features focused on that, like very granular security and governance and data marketplace. There's also some non-enterprise features that ClickHouse lacks, like the ability to execute Python in database (so you can bring ML in).

But the biggest difference is that Snowflake is storage segregated architecture. Scaling Snowflake is done by running "alter warehouse resize" or something. You can also dedicate specific compute slices to specific users and scale them up and down as needed. And this is all managed for you.

If you want to run ClickHouse at scale, you have to run your own k8s, figure out how to manage persistent storage, figure out how to replicate your data, manage cluster replicated tables, etc. Once you outgrow single instance, things get exponentially more difficult - both for the admins and for the users.

Also, while ClickHouse can do joins and is getting better and better optimizer as we speak, and is probably faster than Snowflake for the same money on "single big table analytics" kind of workload, I would expect it to perform much worse in traditional analytics queries, like you would find in TPC-DS.

I wonder if ClickBench needs more data or something.

I just randomly compared ClickHouse, Databend and Doris on c6a.4xlarge machine. For some queries, there's a big differences, like ClickHouse being order of magnitude faster than the others in Q28 and two orders of magnitude slower in Q29. That sounds useful because it shows some of the databases do something differently than the others and areas for improvement.

But for most of the queries, the comparison is like 0.02s vs 0.03s vs 0.03s. That doesn't sound very meaningful and also I wonder how precise the measurement is, since if we're looking at difference of milliseconds it is much easier for randomness to sneak in, compared to measurements in minutes.

While I can read it in a way that modern columnar databases are superfast and awesome, I wonder how the results would look like on order or two orders of magnitude more data.

Clickhouse looks more complicated because instead of

    create table bla (
      id number,
      xyz varchar
    )

you're doing things like

    create table bla (
      id uint64 CODEC(DoubleDelta(8)),
      xyz Nullable(LowCardinality(Varchar)) CODEC(ZSTD(1))
    )
    engine=MergeTree
    order by id

but I do wonder what is the performance difference between actually carefully specifying everything and leaving defaults (except for order by I guess).

> The other thing that rubbed me the wrong way was that rather than fix the issue, they just removed functionality.

Yeah, DuckDB has some very cool features, but I with the community were less abrasive. I remember someone asking for ORC columnar format support, and DuckDB replied "that is not as popular as Parquet so we're not doing it, issue closed". Same story with Delta vs Iceberg.

Meanwhile Clickhouse supports both and if you ask for things they might say "tha tis low priority but we'll take a look". Clickhouse-local can work as CLI (though not in-process) DuckDB too.

As long as Sans is around.

I choose Postgres every time, because that is what I am familiar with.

But nerdy stuff:

Postgres stores data linearly (in heap - which has nothing to do with the heap data structure used for sorting, it just means pile of data). If you need to have fast access to data, you need to add secondary indexes - and the secondary indexes point to location in the heap as "this is where you find the data".

MySQL stores data in a tree - a table is a tree sorted by primary key. You can create secondary indexes and instead of a pointer they contain the primary key value.

That means for example that data with similar primary key will be located physically nearby each other, in MySQL but not in Postgres. At the same time, inserting new data with random (like UUID) primary key in MySQL will write all over the table, but will mostly "append at the end" in Postgres.

Postgres also implements MVCC with Serializable Snapshot Isolation - so data that someone changes exists in multiple copies and needs to be cleaned up later - but there's no locking. MySQL relies on locks instead so there's no duplication but you might see transactions waiting for each other. I don't remember if MySQL implements a proper serializable isolation - but that is not really the default on any database anyway.

Interestingly, Oracle has very similar design to Postgres (though it uses rollback segment for old data, so there's no bloat and vacuum but you might get "snapshot too old" error) while MS SQL Server is also tree and lock-based database like MySQL.

Does this impact you? It might, like in cases where MySQL performs terribly due to UUID keys or Postgres can't vacuum fast enough due to high volume of updates or something. Or you're implementing money settlement logic and need proper serilizable transactions, who know. But it is cool to know the implementation details.

I am a Millennial and I do feel connection to some creators. I am sad that I won't be able to watch the content anymore if Youtube blocks me. I will miss them and I will miss other cool content, like recorded conference talks and tutorial and stuff.

I did try Youtube without adblocker and it was unusable, and Google is horrible evil company that I am for sure not giving any money.

I'll miss it, but at the same time, Youtube is horrible time thief. It is bittersweet but it will make my life better to be without it. If I spent the time I have over the years wasted on Youtube instead reading books, hiking or working out, I'd be a much better person today.

How are degoogled phones nowadays?

Last time I tried that, it was very limited. No paying by phone, no banking apps (and most banks require app as second factor), no proper maps with navigation and routing, no streaming music, no "work" apps like Teams or Slack, no utility apps like taxis or public transport. Even messaging apps wouldn't work well due to lack of push notifications.

Degoogled Android was basically touch display dumbphone. Which is perfect for some people who want it that way, but I like the ability to use public transport, get a cab or message people by something else than a text.

edit: I think there might be a difference between degoogled as in not registered to play store but still having play services install, and degoogled as using shim play services? I honestly don't remember.

> Moss and vines look great on Britain's Georgian and Gothic architecture. On Brutalist flat surfaces, it just looks dirty.

That's just like your opinion, man.

> Brutalism is an academic justification of the brutalisation of the working class, who invariably were the ones who ended up living in cheaply constructed Brutalist architecture. The middle and upper classes (and the architects) keep to their pretty Georgian Neo-classical houses.

That is a projection. Most brutalist buildings are not living spaces but things like bridges, university buildings, libraries, airports, museums, etc. Who do those oppress?

If you don't like the concrete because the Soviets used a lot of it, say that without the pseudo-intelecualization.

It is tricky - on one hand, it seems pretty nice to cover places like car parks or some crops with roofs of solar panels. You get to use the space twice and you get some shade.

On the other hand, the panels are pretty heavy and need to be cleaned every so often. You need a lot of steel to make sturdy framework to put the panels on, unless you want them to fall on someone or break in strong wind. And cleaning the panels that are high up is more difficult than cleaning panels that are ground level.

So depending on cost of labor and cost of steel and whatever (and I don't have numbers, I'm just musing) you might be better of putting solar panels on separate empty field.

"Citron" is how bunch of languages (including mine) refers to lemon. This was pretty confusing article until I figured that out.

Yeah, I always thought datacenters would use Halon. It of course has the problem of suffocating everyone.

Water and fire already had their way, I suspect tornado and a landslide are next.

Yes, but people will still not stop complaining and arguing about all of those.

There is a transaction / amount limit for contactless transactions. Banks generally keep it secret for security - but it can be something like "transactions over x EUR need PIN and also ever fifth will need PIN".

It is pretty annoying because there are a lot of terminals that don't have the PIN keyboard, e.g. on vending machines. If you card decides it needs PIN, you are not buying, unless you can sidestep it using a phone.

You forgot the people who love SQL and people who hate SQL.

Indeed. The evil always wins in the end, such is the nature of the world.

We are talking about genocidal war where the first thing the occupiers establish in occupied territories are torture rooms. Just last week, Russian soldiers admited on camera they had orders to murder children and went along with it.

What kind of self reflection would make people who are ok with that not complete filth I wonder?

I know some people want Russia to win, and they are real. They are either ignorant or they are horrible and evil people.

I also wonder why what is basically Russian propaganda account has no limits and how it got to 8k karma, when it has basically nothing to say.

I know there is practice of buying established accounts with bunch of karma / followers to use for both marketing and propaganda purposes so that might be this case.

Wouldn't architect role be something you're looking for? Of course, there's a lot of different types of architects and some are just bullshit slidemakers, but you could get some coding and a lot of complex problem solving in right architect role.

I think that goes in general. White collar crime pays, most if it goes uncovered and when it does, you might get few years of prison, while still ending up being rich for rest of your life.

Few years of 24/7 prison or 30+ years of 9-5 prison?

I looked it up, and it's "content management company" Little Dot Studios, who seem to make money by claiming things on youtube to steal ad money and extort people.

> Video unavailable. This video contains content from LDS, who has blocked it in your country on copyright grounds

I hope LDS burns in hell, whoever they are.

Parquet is not really a new format - it has been around for years and years in the Hadoop world. It is mostly good for storing large data in a way that makes it effective to process it for OLAP workload, and that is it.

You can give them anything from SQL interface to BI tool like Superset to no-code tool like Thoughtspot. They might still export some to CSV, but having GUI is much better than having CSV.

Parquet doesn't really do timestamp with time zone. Also relevantly for processing Parquet, Apache Spark doesn't really do timestamp with time zone. Meaning that if your data go through parquet (e.g. in a data lake) you have to store timestamps as strings, or lose the time zone.

It sucks.

That however only impacts CSVs or other text files, where the compression is the envelope. In parquet, it is the internal chunks that are compressed, so it is as parallelizable as ever.

Also, generic CSV is not parallelizable anyway, because it allows you to have (quoted) newlines inside fields and then there is really now way to split it on newlines because you never know if you're on beginning or row or in the middle.

So parallel reading of CSVs is more of a optimization for specific CSVs that you know don't have newlines inside fields.

Don't do it. I bought M1 Air instead of M1 14" Pro to save money and now I regret it. The Pro's display and speakers are much better I now watch movies on work M1 14" Pro.

M2 Air might be good too, I don't know. But if I could go back, I'd go to M1 Pro.

Sometimes, people just say things that are utterly vile and disgusting, and then they get downvoted.

> Ukraine may win... after 5+ years of brutal war and heavy destruction. Is that better than a life back under Moscow's rule?

Hey that's a good example!