Ask HN: 哪个同步引擎的模块化程度最高？
1 分•作者: olivieropinotti•6 个月前
嗨，HN！
我正在开发一款新应用，希望它能像 Linear 那样快速响应，实现实时查询、同步和写入操作。
我对这方面并不精通，欢迎大家指教。我研究过一些方案，比如 Convex、ElectricSQL、Zero、Liveblocks 等。
我觉得它们总是在某些方面有所欠缺，最能概括这一点的就是它们的模块化。
我理想中的同步引擎应该具备以下特点：
* 能够使用我的数据库（Postgres，AWS RDS）
* 能够在我的后端构建和执行查询（假设我的前端是 Next.js (Vercel)，有一个 API 路由指向我的 FastAPI 服务器 (AWS ECS)，我在那里处理所有身份验证/权限中间件等）
* 有一种简单且熟悉的方式来声明同步引擎的模式（比如重用 SQLAlchemy 或 Drizzle 的模式）
* 一个使用 SQL 的简单 SDK 来构建查询
此外，我还想知道这样的系统如何处理连接池、分片、复制等问题。
有类似的东西存在吗？或者是什么主要挑战阻碍了它的出现？
查看原文
Hey HN!<p>For a new app I'm developing I'd like to give it the fast and responsive feel that apps like Linear have. By having real-time querying, syncing and write operations.<p>I'm no expert on the matter, so feel free to educate me. I have looked into a couple of options such as Convex, ElectricSQL, Zero, Liveblocks, etc..<p>I feel like they always come short in a couple of ways and the best way to summarize that is their modularity.<p>What I would love in a sync engine:
- Being able to use my database (Postgres, AWS RDS)
- Being able to form and execute queries in my backend (suppose I have frontend in Next.js (Vercel) with an API route to my FastAPI server (AWS ECS) where I have all of my auth / permission middleware, etc.)
- Have a simple and familiar way to declare the schemas to the sync engine (like re-using SQLalchemy's or Drizzle's schemas)
- A simple SDK to form queries that uses SQL<p>Further, I wonder how such a system would work with connection pooling, sharding, replication etc.<p>Does something like this exist? Or what are the major challenges that prevent this from existing?