H2 Database by Hypersonic Creator Nearing 1.0
v0.9 addresses bug fixes and also adds CSV import/export, the REAL data type is supported, better performance for joins, access rights for views, and invalid views (if a table is dropped).
H2 was first announced in mid-december 2005 and work has progressed steadily since. Thomas' priority for H2 is currently compatibility and performance on the low end (single user, Access, Embedded), so features such as better transaction isolation support (it currently only supports serializable) will be implemented later.
Comparison to Other Database Engines
|Embedded Mode (Java)||Yes||Yes||Yes||Yes||No||No|
|Cost Based Optimizer||Yes||No||Yes||Yes||Yes||Yes|
|Files per Database||Few||Few||Many||Few||Many||Many|
|Footprint (jar/dll size)||~ 1 MB||~ 600 KB||~ 2 MB||~ 3 MB||~ 4 MB||~ 6 MB|
H2 has also benchmarked sigificantly faster than other open source databases, including HSQLDB:
According to Thomas, H2 may be merged back to HSQLDB at some point in the future, but not yet, which is one reason why the license is not GPL / LGPL / Apache style, although it is free and open source.
Native vs. Java versions
Any comparisons between those versions?
Also, locking modes
IMHO the table level locking for managing concurent access is quite a problem for multiple users scenarios. May be DB2 is slower but IBM uses row based locking, a policy that can give you more throughput in multiple users usecases.
Re: Also, locking modes
Read carefully the "Locking, Lock-Timeout, Deadlocks" section from www.h2database.com/html/frame.html IMHO the table level locking for managing concurent access is quite a problem for multiple users scenarios. May be DB2 is slower but IBM uses row based locking, a policy that can give you more throughput in multiple users usecases.
True, but as mentioned in the article, Thomas' first priority is making the DB work best "on the low end (single user, Access, Embedded)". I think this is the largest use scenario for HSQLDB and H2. Here at InfoQ we used HSQLDB on developer dev machines (single user, embedded) to simplify the environment instead of the production DB.
Free and open source.
(I'm not an open source zealot, but I am an accuracy zealot)
From the FAQ:
Is this Database Engine Open Source?
It is free to use and distribute, and the source code is included. But currently does not have a 'open source style license', but this will probably change soon. See also under license.
Re: Native vs. Java versions
The GCJ version is not as stable as the Java version.
Currently, the GCJ version is also slower than when using the Sun VM.