The Rise of the Distributed SQL Database

crate.io

7 points·nslater··5 comments

This focuses on open source databases. Distributed SQL databases go clear back to the 1980s.

Do you have an example of one? I don't think anything from the pre-CAP time can call itself distributed database, i.e. anything from the 1980s or 1990s or even early 2000s. They usually made an assumption that the network is fast and reliable, essentially ignoring all of the problems post-CAP distributed databases address.

Informix did. Perhaps Ingres or Oracle, also.

Hmm, two-phase commit Informix included in like late 90s, which is pretty much the only thing related to distributed systems it had. None of them were really distributed databases in any way of course. The closest thing to a distributed database you could have back then was NonStop SQL, but it was designed with an assumption of a fast reliable network or more like a supercomputer.

I guess I was imagining it when I used the features myself internally starting in 1989 or so.

Distributed client/server was first implemented in Informix-SE in the late 1980s but was more widely used starting with Informix-OnLine. Version 5.0 (general release in late 1991) had two-phased commit, but I think 4.X did also no later than summer 1990 because I did some onsite diagnosis/fixing for a major client then.

There are references to OnLine 5.0 features and general release in the Informix 1991 Annual Report[0].

[0] http://archive.computerhistory.org/resources/text/Informix/i...

Reply on news.ycombinator.com