Skip to content

Add Turso (it is unbelievably slow) (it also does not work) #336

Open

Description

@alexey-milovidov

It should be a drop-in replacement for SQLite. It is written in Rust.

Activity

  1. changed the title [-]Add Limbo[/-] [+]Add Turso[/+] on Jul 11, 2025
  2. changed the title [-]Add Turso[/-] [+]Add Turso (but it's difficult because it panics in Rust)[/+] on Jul 12, 2025
  3. changed the title [-]Add Turso (but it's difficult because it panics in Rust)[/-] [+]Add Turso[/+] on Jul 12, 2025
  4. alexey-milovidov commented on Jul 12, 2025

    @alexey-milovidov
    MemberAuthor
  5. changed the title [-]Add Turso[/-] [+]Add Turso (but it's difficult because it does not load the data)[/+] on Jul 13, 2025
  6. alexey-milovidov commented on Jul 13, 2025

    @alexey-milovidov
    MemberAuthor
  7. alexey-milovidov commented on Jul 21, 2025

    @alexey-milovidov
    MemberAuthor

    It was loading the data for a week already, and the speed of loading data has dropped to 4 kilobytes 🤣 per second, and it will take years to load the dataset.

  8. alexey-milovidov commented on Aug 2, 2025

    @alexey-milovidov
    MemberAuthor

    It loaded maybe 1% of the data so far.

  9. alexey-milovidov commented on Aug 17, 2025

    @alexey-milovidov
    MemberAuthor

    The speed has lowered to about 1 KB per second, which is possible if the database is generated by LLM, and not the best one.

  10. changed the title [-]Add Turso (but it's difficult because it does not load the data)[/-] [+]Add Turso (it is unbelievably slow)[/+] on Aug 17, 2025
  11. changed the title [-]Add Turso (it is unbelievably slow)[/-] [+]Add Turso (it is unbelievably slow) (it also does not work)[/+] on May 12, 2026
  12. alexey-milovidov commented on May 12, 2026

    @alexey-milovidov
    MemberAuthor

    I didn't load the data after a year.

    I tried it with the fsync removed, but even then it does not work:

      Turso — 3 distinct runs:
    
      time: 2026-05-11 21:54:56
      machine: c6a.4xlarge
      outcome: load killed by SIGTERM at ~6353 s (run timeout was set to 864000 s, so likely earlyoom on the tursodb import); mydb ended up at 8192 bytes (empty SQLite header). All 43 [t1,t2,t3], rows emitted but
        timings are essentially 0 since the table has 0 rows — bogus result, must not be committed.
    
  13. alexey-milovidov commented on May 14, 2026

    @alexey-milovidov
    MemberAuthor

    Now it aborts with Rust's panic:

    Turso v0.1.2-pre.4
    Enter ".help" for usage hints.
    This software is ALPHA, only use for development, testing, and experimentation.
    thread \'main\' panicked at core/storage/sqlite3_ondisk.rs:818:5:
    assertion failed: !*syncing.borrow()
    note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
    /opt/clickbench/system/load: line 12:  1162 Aborted                 (core dumped) tursodb mydb \'.import --csv hits.csv hits\'
    === load done rc=134 in 438.4s ===
    
  14. Pavan-Nambi commented on May 18, 2026

    @Pavan-Nambi

    Hi, sorry about this, most of the time now focus was on correctness. but..

    Turso v0.1.2-pre.4

    hmm maybe i am misunderstanding something or is this just stale print? turso new release is 0.6 , 0.1.2 is like way too old... like almost a year old...

  15. Pavan-Nambi commented on May 18, 2026

    @Pavan-Nambi

    i spinned an ec2 instance(c6a.4xlarge, 32gig, 16vcpus) for this

    with v 0.6 it's neither panicking nor terminating for me.

    i'm not sure if i am maybe doing something wrong, could u please try with v0.6?

    i stopped at query 2 , 3rd try

      {
        "system": "Turso",
        "date": "2026-05-18",
        "machine": "c6a.4xlarge",
        "cluster_size": 1,
        "proprietary": "no",
        "hardware": "cpu",
        "tuned": "no",
        "tags": ["Rust", "embedded", "row-oriented"],
        "load_time": 6564.021,
        "data_size": 68702932992,
        "result": [
          [318.640, 305.446, 305.469],
          [337.962, 325.489, null]
        ]
      }
    
    
  16. alexey-milovidov commented on Jul 16, 2026

    @alexey-milovidov
    MemberAuthor

    Yes, I've started this load about a year ago, and it didn't finish. Will try the new version.

  17. Pavan-Nambi commented on Jul 16, 2026

    @Pavan-Nambi

    (if you are trying new version - we released 0.7 few days ago.)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions