They have very good goals but their new approach to have database files memory-mapped gave me the chills.
Ups:
* No in-process cache required.
* Very clean design.
Downs:
* Every read from disk would bring a lot more data than required.
* 64-bit process required to support large files.
* Since they use a JavaScript VM this would consume 40-60% more memory
because of double pointer size.
* V8 is x86 only so they switched to SpiderMonkey
(slower, even with upcomming tracing.)
The project looks very promising and the developers are top-notch but IMHO it's very green and small for my project. Maybe the next major release will be a killer.
I am not any kind of authority in this area so please take this comment with a grain of salt.
Ups:
Downs: The project looks very promising and the developers are top-notch but IMHO it's very green and small for my project. Maybe the next major release will be a killer.I am not any kind of authority in this area so please take this comment with a grain of salt.