Georg Traar

@arvidkahl Maybe you want to give @cratedb a try? Lucene base full text search / SQL / Postgres Wire Protocol / Open Source (Apache 2.0)

Full-text search on 500GB+ of data is keeping me awake at night. MySQL's full-text index just can't handle this. Often takes minutes. And Meilisearch, as fast as it is, is hard to wrangle to get it to get only precise results. Anyone here experienced with search at this size?

@jamessewell @samokhvalov @PostgreSQL @TimescaleDB It's misleading marketing to suggest that data is stored in Postgres when, in fact, the bulk of it is kept as Parquet files on S3. Charging for an imaginary Postgres „data size“ is the cherry on top.

@getraar @samokhvalov @PostgreSQL @TimescaleDB So you're worried about the size on disk vs the size of the data? IDK, seems like you're trying to force an argument here.

This is VERY impressive: 1 PiB @PostgreSQL database on a single moderately sized VM in AWS. If I understand correctly, this instance uses a component that is available in @TimescaleDB Cloud and allows storing data of specific partitions in cloud object storage (S3 in this case) transparently. Everyone's dream -- offload old partitions to some long-term cheaper storage. Unlocking potentially unlimited growth: as I remember, both on AWS and GCP, max EBS / SSD PD size is 64 TiB but you can attach multiple volumes combining them using LVM2 or madm, unlocking further PGDATA growth, but still very limited. E.g., on GCP, it's 512 TiB per VM maximum. With the great tech from Timescale, this limit is solved. But it also means, unfortunately, it's not pure @PostgreSQL already, because this component is not open source, I cannot use it not being paying customer of Timescale Cloud, cannot use it free anywhere and manage. Arbitrary Postgres cluster cannot have this yet. One day, maybe.

@jamessewell @samokhvalov @PostgreSQL @TimescaleDB And in terms of charging for "theoretical size". This is from your blog post on tiered storage. timescale.com/blog/scaling-p…

@jamessewell @samokhvalov @PostgreSQL @TimescaleDB How much of the 350TB is stored in PostgreSQL tables after 20x compression, and how much is stored on S3 as Parquet files (250TB mentioned in a blog post)? In an earlier post, you stated that 130TB are 5TB PostgreSQL tables, which would indicate a DB size of 5TB, not 130TB.

@stefano_mancini @MongoDB @TimescaleDB @InfluxDB @cratedb might be worth a look as you can combine the ability of working with documents, while still having columnar storage

What’s your preferred time series database? And why you chose it? My main application database is @MongoDB , and it supports a light time series implementation, but I’m looking for alternatives. I’m the past I used both @TimescaleDB and @InfluxDB with nice results.

@tomasmcguinness If it is just about a single reader/writer you might want to look at @DuckDB (analytics eqv. of SQLite) If you want to monitor 1000s and provide dashboards to 100s, you might want to look at @CrateDB 😉

For my radiator temperature sensor recoding, I originally planned on just using SQLite, but I wonder if InfluxDB wouldn't be better as it's time series data.

@mfussenegger complex docs systems with brittle build processes probably don’t help with later.

@DavidSoles_ @mfussenegger CrateDB + crash 🫶

@mfussenegger 🤣

@InfluxDB Yes, we (@cratedb) can. with dynamic mapping and evolving schemas

You can't create a #SQL database without a predefined schema, where the data is normalized into tables with primary and foreign keys designated. Learn more: bit.ly/3t65Cfi #NoSQL #InfluxDB

@mfussenegger The biggest differences are the large casting machines, which are unique to Tesla. And typically you would not even be allowed to bring any camera into an automotive assembly line.

@criccomini @PrefectIO We did it last year with @cratedb 🙃

Remarkable to see @PrefectIO go all-in on Apache 2.0 when others are moving the opposite direction.

@jodok @mfussenegger Money buys stake in blockchains as well. Blockchains are not as „decentralized“ as they would like to be. > 75% of Bitcoin mining is handled by a few mining pools in china.

@mfussenegger one thing (citing from the article i shared above): "And yet it has one dimension along which it is radically different. No single entity or small group of entities controls it – something people try to convey, albeit poorly, by saying it is “decentralized.”"

why web3/crypto matters. spoiler: i'm brewing something in there :)

@mfussenegger Pretty sure Visual Studio Code is a Web Browser 😶

@corck Its depressing to read ppl linking papers trying to spin a story around them by adding a lot off “might” and uncertainty, while they don’t even understand the content at all.

Its depressing to read ppl citing journalists argumenting pro vaccination of all groups when biochemist release papers stating that that these experimental vaccines *might* impede DNA repair functionality. The fronts cant be further apart unfortunately. mdpi.com/1999-4915/13/1…

@opennima @cratedb We will foucs on more technical content at @cratedb, while @crateio focuses more on the company in general and other products. For discussions we also have our discourse at community.crate.io

@cratedb Wo?

Welcome to the new @cratedb community page 🥳 Here you will find useful information about #CrateDB, our technical blog posts, documentation, and updates on the releases of new versions. Questions? Feedback? Give a shout!😉

@mfussenegger There is no such a thing a stranded / free energy. All energy is lost as heat, the exergy approaches 0. Most crypto currencies are terrible in terms of sustainability. But for greed all nature is to little.

@andreineculau @TimescaleDB @acoustik I think a problem with (database) benchmarks is, that they are always biased … and that is ok. If you decide which db technologies to use on 50% higher ingest numbers or 50% less storage used then you are probably doing it wrong.

@chaudum Kann man sich sparen … „Die Sozialdemokraten sind schuld“ und ein Haufen „Mimimi“