[Cosmo-dev] cosmo - jcr

[Brian Moseley]

On 9/15/06, Daniel Gradecak <daniel at gradecak.com> wrote:
> In fact I would like to know the reason why cosmo is no more using JCR
> for storage. Is that possible?

we made the change because, although jcr is an attractive api for
storing our data, the implementation we chose - jackrabbit - is not
mature enough for the scalability we require for cosmo. jackrabbit
lacks capabilities like in-place schema upgrades and online
backup/restore, is not clusterable, and is not practically remotely
deployable yet. other vendors may have offered some of these features,
but we chose instead to go with technology (hibernate and rdbms) that
the team has experience scaling.