zX41ZdbW

42 ms·

1. 661.▲
   by zX41ZdbW 4y ago

   It is not available as a library, only CLI.
2. 662.▲
   by zX41ZdbW 4y ago

   Yes, it is detecting curl by the user agent - by using the "page rules" in Cloudflare. Cloudflare is good for traffic management rules. I've created this distribution channel as an experiment and for my own need - I use it to
3. 663.▲
   by zX41ZdbW 4y ago

   Here is one example: https://colab.research.google.com/github/dcmoura/spyql/blob/... Note: DuckDB is not present in this comparison, because it does not support JSON processing (only CSV and Parquet). Cl
4. 664.▲
   Show HN: ClickHouse-local – a small tool for serverless data analytics(clickhouse.com)

   289 points by zX41ZdbW 4y ago|70 comments
5. 665.▲
   by zX41ZdbW 4y ago

   Here is the benchmark comparing all generation of Graviton servers in AWS with Intel and AMD servers: https://benchmark.clickhouse.com/hardware/
6. 666.▲
   by zX41ZdbW 4y ago

   It can be about the memory bandwidth, but it matters mostly for hi-end data processing software. You can compare Threadripper (4-channel memory), Threadripper Pro (8-channel memory, same as for EPYC servers) and M1 by memory bandwidth. Exam
7. 667.▲
   by zX41ZdbW 4y ago

   Cross-compiling is more convenient. If you have an amd64 desktop or aarch64 server, you can compile to amd64 and aarch64, and powerpc64 with almost the same command. The binary without dependencies can be copied to the server with `scp`. Th
8. 668.▲
   by zX41ZdbW 4y ago

   Databend is one of a few DBMS modeled directly after ClickHouse. It currently does not approach a similar performance [1], and the RRIR motivation is dubious [2]. [1] https://benchmark.clickhouse.com/ [2] https:/&#x2F
9. 669.▲
   by zX41ZdbW 4y ago

   Touché.
10. 670.▲
    by zX41ZdbW 4y ago

    I want to ask a question to the engineers with experience in WebAssembly. Last year I invited a student to compile ClickHouse with WebAssembly. But the task appeared harder than expected, supposedly due to poor 64-bit support in WASM. Here
11. 671.▲
    by zX41ZdbW 4y ago

    SingleStore is a kind of uninteresting example because it is not open-source. It makes potential customers feel non-confident. Why move large data volumes to SingleStore if the company can cease to exist in a year?
12. 672.▲
    by zX41ZdbW 4y ago

    ClickBench is still ok to compare non-OLAP databases on OLAP workloads - filter away the column-oriented DBMS, and you will be left with the comparison of non-OLAP databases. I think it makes sense to add NeonDB for the sake of completeness
13. 673.▲
    by zX41ZdbW 4y ago

    What I found strange about MariaDB is that it is ~10 times slower than MySQL on OLAP queries[1]. It looks entirely wrong and most likely related to an issue with Debian packages. [1] https://pastila.nl/?00697bd3/ef6bf88
14. 674.▲
    by zX41ZdbW 4y ago

    You can use ClickHouse as a stateless query engine to process externally hosted datasets. This way you just publish a MergeTree table to S3 bucket or a set of static files on an HTTP server, and then plugging it on an empty ClickHouse serve
15. 675.▲
    by zX41ZdbW 4y ago

    Let's commit this to ClickBench: https://github.com/neondatabase/neon/pull/2855 So, NeonDB can be compared with other serverless Postgres offerings.
16. 676.▲
    by zX41ZdbW 4y ago

    > small data. The biggest winner in this domain is Motherduck I think ClickHouse is a perfect choice for "small data". Easy to install, self-contained, portable... but not as limiting as DuckDB.
17. 677.▲
    by zX41ZdbW 4y ago

    ClickHouse is the one that is suitable for a wide range of applications. See: https://clickhouse.com/docs/en/about-us/adopters/ The main advantage of ClickHouse is that it's a normal SQL relational
18. 678.▲
    by zX41ZdbW 4y ago

    > e.g. no CTEs in ClickHouse ClickHouse has support for CTE. For example, this HTML page I've created for work is using CTEs to query ClickHouse, check out the view-source: https://aretestsgreenyet.com/
19. 679.▲
    by zX41ZdbW 4y ago

    It is even worse from the usability standpoint. Compare: https://play.clickhouse.com/play?user=play and https://github.demo.trial.altinity.cloud:8443/play?user=demo (password demo) Details matter.
20. 680.▲
    by zX41ZdbW 4y ago

    @hodgesrm it's simply a copy of the Playground I've created in 2020. There is nothing wrong with copycats because we allow them. But nothing to be proud of from your side. It's also working slower, has less number of datasets
21. 681.▲
    by zX41ZdbW 4y ago

    The most developed analytical query engine as of today is ClickHouse [1]. It is open-source, has the separation of storage and compute and the best performance. ClickHouse is a source of inspiration for many emerging database engines: Velox
22. 682.▲
    by zX41ZdbW 4y ago

    > he claimed that because MonetDB focuses on read-only OLAP workloads, then MMAP was good enough ClickHouse has switchable IO engines: - read; - pread; - mmap; - pread_threadpool; The best performance is achieved with the combine
23. 683.▲
    by zX41ZdbW 4y ago

    Here is a dashboard of the Playground server utilization: https://pastila.nl/?00be67ef/eaf1c2bc2f6edc620741ff456c9f493... The dashboard is also implemented with ClickHouse, as single HTML page. Source code: https:&#x2
24. 684.▲
    by zX41ZdbW 4y ago

    It does not handle deletes / updates of the source table at all - it means the data of the materialized view become unaffected. Nothing sophisticated... About the multi-level aggregate queries - the only way is to define a materialized
25. 685.▲
    by zX41ZdbW 4y ago

    ClickHouse updating materialized views in realtime. How it is done: 1. Intermediate states of aggregate functions are first-class data types in ClickHouse. You can create a data type for the intermediate state of count(distinct), or quantil
26. 686.▲
    by zX41ZdbW 4y ago

    Yes, it's continuously updated. The source code is here: https://github.com/ClickHouse/github-explorer This shell scripts updates it: https://github.com/ClickHouse/github-explorer/blob&#x
27. 687.▲
    by zX41ZdbW 4y ago

    Yes, the native format is very similar to Parquet. It works for Parquet as well: SELECT * FROM url('https://clickhouse-public-datasets.s3.amazonaws.com/hits.parquet') LIMIT 1 And for CSV or TSV: SELECT *
28. 688.▲
    by zX41ZdbW 4y ago

    clickhouse-client Download it as: curl https://clickhouse.com/ | sh Connect to the demo service: clickhouse-client --host play.clickhouse.com --user play --secure
29. 689.▲
    by zX41ZdbW 4y ago

    The cost can be calculated, but not always. Example 1: SELECT ... FROM table WHERE key BETWEEN c1 AND c2 - here we can get the number of records and bytes to scan before even starting the query. Example 2: SELECT ... FROM table W
30. 690.▲
    by zX41ZdbW 4y ago

    I have prepared a nice report: the rank of the websites in groups 1..10, 11..100, ... the percentage of TLS and an example of non-TLS website: https://play.clickhouse.com/play?user=play#U0VMRUNUIGZsb29yK... SELECT