We are using successfully H2 1.4.x in production for almost 2 years in a big organization with 300 of users. It's an embedded database behind GIS desktop system which is a main tool these 300 users are using daily. So, very business critical case :) Data (spatial data) is taken from central database (Postgre) and delivered to H2 where all kind of CRUD operations are performed. Changes are collected and synchronized back to central database.
We always compile H2 from code with minor local changes optimizing the performance for certain SQL scenarios (probably some improvements are worth to contribute to master branch). PageStore mechanism is used. So far the biggest problem was the performance and stability of BLOB/CLOB types. Since we got rid of them towards VARBINARY/VARCHAR types the number of corrupted database cases dropped to almost 0. Actually it was mistake to use BLOB in datamodel design from the beginning, once I have learned better H2 internals an understanding that BLOB is not needed in our case solved the majority of database corruption cases from users. There is so much to do still in H2, but it's great Java product, thanks to developers! P.S. Our optimizations concerned nested selects like .... WHERE FK_COLUMN IN (SELECT... ) . Often when index exists for FK_COLUMN, parent table contains 200000 records and nested SELECT retrieves thousands of values implementation is not optimal; I try to write soon about this case with a patch suggested. -- You received this message because you are subscribed to the Google Groups "H2 Database" group. To unsubscribe from this group and stop receiving emails from it, send an email to h2-database+unsubscr...@googlegroups.com. To post to this group, send email to h2-database@googlegroups.com. Visit this group at https://groups.google.com/group/h2-database. For more options, visit https://groups.google.com/d/optout.
