Hacker News Reader

Comment by beardicus | Hacker News Reader

> LiteTree is more than TWICE AS FAST than normal SQLite on Linux and MacOSX!!!

In my experience, claims like these usually end up showing that the author didn't understand the `PRAGMA synchronous` setting at all, or they chose to ignore it to juice their stats.

In this benchmarking test are the data durability guarantees the same for both LiteTree and vanilla SQLite?

View on HN

I also read this sentence and my immediate thought was that I would only use three exclamation marks in the documentation for a database engine to say something like "TWICE AS MUCH test coverage!!!"

If I want a fast database, I will just write my data to /dev/null. It has well-understood data durability guarantees and is very quick to write to.

I prefer /dev/urandom or /dev/zero. They have a much higher likelihood of returning the data I wrote.

True! For maximum I/O efficiency, obviously you write to /dev/null and read from /dev/urandom.

Just how I like my CQRS!

No, no, /dev/urandom is really slow...

> In this benchmarking test are the data durability guarantees the same for both LiteTree and vanilla SQLite?

Who cares if you get more bongoliomarks.

Looks like the speed comes from using lmdb rather than sqlites own storage layer.

I first read this as IMDd for Internet Movie Database and got confused, but probably you mean LMDB for Lightning Memory-Mapped Database, which makes much more sense. :)

lmdb != Imdb

Check your fonts if those two look the same.

But, AFAIK, lmdb is not write optimized (as opposed to something that uses LSM trees like levelDB).

I wonder if they got the inspiration from https://github.com/LMDB/sqlightning

Reply on news.ycombinator.com