4 ms·

By the way, the emergence of LLM coding tools could make it even easier than before to reduce that dependence, as the cost of reproducing many of the mature tec

By the way, the emergence of LLM coding tools could make it even easier than before to reduce that dependence, as the cost of reproducing many of the mature technologies is going to cost less than it would have before. Ironically, doing that may require using US tools (like Claude Code), at least for now, but it could be a very interesting evolution/opportunity for Europe.

petcat 9mo ago
> the emergence of LLM coding tools could make it even easier than before I find this highly optimistic. It will take years, maybe decades for EU to replace US clouds and tech. And if they're going to do it with LLMs, then it will take billions of euros in devs and tokens (again, all going to US tech companies). Meanwhile, USA continues to strategically re-home TSMC to Arizona whilst simultaneously make huge investments to invigorate Intel and Micron. Over the last decade USA and China have doubled-down on massive investments to out-compete each other while the EU seems like it's struggling to understand where to even begin.
- johnnyanmac 9mo ago
  > USA continues to strategically re-home TSMC to Arizona whilst simultaneously make huge investments to invigorate Intel and Micron. Oh don't worry, Trump's already kneecapped both of those for a decade to come from 2025's actions alone. Y'all got time to catch up. China, much scarier. But we all kinda let that happen over 30 years. Too late to complain now. I'd say we work together but uhh... I think we both understand (or rather, fail to understand) modern US policy these days.
raw_anon_1111 9mo ago
Yes, I can see Claude Code making it easier to reproduce - Redshift (or Snowflake) - or anything else you need to be reliable and performant at scale.
- vjerancrnjak 9mo ago
  Both products are nothing but reliable. Redshift can’t even go around partitioning limits, or S3 limits. But what’s funny is that Claude Code is from US company so can’t be used in a boycott scenario
  - raw_anon_1111 9mo ago
    Redshift is used at the largest e-commerce site in the world and was built specifically to “shift” away from “Big Red” (Oracle).
    - vjerancrnjak 9mo ago
      What can I say, I expected more than what they actually offer. A Redshift job can fail because S3 tells it to slow down. How can I make this HA performance product slower given its whole moat is an S3 based input output interface. As a compute engine its SQL capabilities are worse than the slowest pretend timeseries db like Elasticsearch.
      - raw_anon_1111 9mo ago
        Are you trying to treat an OLAP database with columnar storage like an OLTP database? If you are, you would probably have the same issue with Snowflake. As far as S3, are you trying to ingest a lot of small files or one large file? Again Redshift is optimized for bulk imports.
        vjerancrnjak 9mo ago
        Redshift does not fit into aws ecosystem. If you use kinesis, you get up to 500 paritions with a bunch of tiny files, now I have to build a pipeline after kinesis that puts all of it into 1 s3 file, only to then import it into redshift which might again put it on s3 backed storage for Its own file shenanigans. Clickhouse, even chdb inmemory magic has better S3 consumer than Redshift. It sucks up those Kinesis files like nothing. Its a mess. Not to mention none of its Column optimizations work and the data footprint of gapless timestamp columns is not basically 0 as it is in any serious OLAP but it is massive, so the way to improve performance is to Just align everything on the same timeline so its computation engine does not beed to figure out how to join stuff that is Actually time Aligned I really can’t figure out how anyone can do seriously big computations with Redshift. Maybe people like waiting hours for their SQL to execute and think software is just that slow.

petcat 9mo ago

> the emergence of LLM coding tools could make it even easier than before I find this highly optimistic. It will take years, maybe decades for EU to replace US clouds and tech. And if they're going to do it with LLMs, then it will take billions of euros in devs and tokens (again, all going to US tech companies). Meanwhile, USA continues to strategically re-home TSMC to Arizona whilst simultaneously make huge investments to invigorate Intel and Micron. Over the last decade USA and China have doubled-down on massive investments to out-compete each other while the EU seems like it's struggling to understand where to even begin.

johnnyanmac 9mo ago
> USA continues to strategically re-home TSMC to Arizona whilst simultaneously make huge investments to invigorate Intel and Micron. Oh don't worry, Trump's already kneecapped both of those for a decade to come from 2025's actions alone. Y'all got time to catch up. China, much scarier. But we all kinda let that happen over 30 years. Too late to complain now. I'd say we work together but uhh... I think we both understand (or rather, fail to understand) modern US policy these days.

raw_anon_1111 9mo ago

Yes, I can see Claude Code making it easier to reproduce - Redshift (or Snowflake) - or anything else you need to be reliable and performant at scale.

vjerancrnjak 9mo ago
Both products are nothing but reliable. Redshift can’t even go around partitioning limits, or S3 limits. But what’s funny is that Claude Code is from US company so can’t be used in a boycott scenario
- raw_anon_1111 9mo ago
  Redshift is used at the largest e-commerce site in the world and was built specifically to “shift” away from “Big Red” (Oracle).
  - vjerancrnjak 9mo ago
    What can I say, I expected more than what they actually offer. A Redshift job can fail because S3 tells it to slow down. How can I make this HA performance product slower given its whole moat is an S3 based input output interface. As a compute engine its SQL capabilities are worse than the slowest pretend timeseries db like Elasticsearch.
    - raw_anon_1111 9mo ago
      Are you trying to treat an OLAP database with columnar storage like an OLTP database? If you are, you would probably have the same issue with Snowflake. As far as S3, are you trying to ingest a lot of small files or one large file? Again Redshift is optimized for bulk imports.
      - vjerancrnjak 9mo ago
        Redshift does not fit into aws ecosystem. If you use kinesis, you get up to 500 paritions with a bunch of tiny files, now I have to build a pipeline after kinesis that puts all of it into 1 s3 file, only to then import it into redshift which might again put it on s3 backed storage for Its own file shenanigans. Clickhouse, even chdb inmemory magic has better S3 consumer than Redshift. It sucks up those Kinesis files like nothing. Its a mess. Not to mention none of its Column optimizations work and the data footprint of gapless timestamp columns is not basically 0 as it is in any serious OLAP but it is massive, so the way to improve performance is to Just align everything on the same timeline so its computation engine does not beed to figure out how to join stuff that is Actually time Aligned I really can’t figure out how anyone can do seriously big computations with Redshift. Maybe people like waiting hours for their SQL to execute and think software is just that slow.