←back to thread

Zero-latency SQLite storage in every Durable Object

(simonwillison.net)
285 points ajhit406 | 1 comments | 13 Oct 24 23:07 UTC | HN request time: 0.21s | source
Show context
tmikaeld ◴[14 Oct 24 06:07 UTC] No.41834673[source]▶
> ..each DO constantly streams a sequence of WAL entries to object storage - batched every 16MB or every ten seconds.

Which also means it may take 10 seconds before you can (reliably) read the write globally.

I keep failing to see how this can replace regionally placed database clusters which can serve a continent in milliseconds.

Edit: I know it uses streams, but those are only to 5 followers and CF have hundreds of datacenters. There is no physical way to guarantee reads in seconds unless all instances of the SQLite are always connected and even then, packet latency will cause issues.

replies(4): >>41834790 #>>41834832 #>>41835155 #>>41840319 #
1. kentonv ◴[14 Oct 24 18:21 UTC] No.41840319[source]▶
As others have noted, you misunderstand how Durable Objects work. All traffic addressed to the same object is routed to a single machine where that object lives. That machine always has a consistent view of its SQLite database. You can have billions of objects, but each has its own separate database. There's no way to read from a database directly from a different machine than the one the DO is running on.