HackerNews中文版

我们正在使用 Dagster 运行 dbt，并在 8 个月内增长到大约 300 个模型。编译时间已经增加到 30 秒以上，我预见到问题的严重性。如果我们按照同样的轨迹发展到 3000 个模型（我们将在 2-3 年内达到），那么编译时间将达到 5 分钟以上。有没有其他人遇到过 dbt 的这种规模瓶颈？在模型数量达到多少时，编译时间变得难以忍受？很想听听大家的经验之谈，或者其他人是否会觉得更快的替代方案有用。

查看原文

We're running dbt with Dagster and have grown to ~300 models in 8 months. Compilation time has crept up to 30+ seconds, and I can see the writing on the wall.<p>If we follow the same trajectory to 3000 models (which we'll hit in 2-3 years), we're looking at 5+ minute compilation times.<p>Has anyone else hit this scaling wall with dbt? At what model count did compilation times become painful? Would love to hear war stories or if others would find a faster alternative useful.

应该用 Rust 重写 dbt。