Hello everyone
We are trying to adopt MongoDB basically for its schema-less nature.
In other words, we have to persist objects with dynamic-structures and
then able to querying for its attributes. Currently, our datasource is
DB2, and we have to persist millions of schema-less-objects (aprox 10
millions per month).
So, we have two alternatives:
- to use MongoDB (or another document-based DB)
- to use the SQL/XML feature of DB2 (https://www.ibm.com/developerworks/d...-0603saracco2/)

Both alternatives have pros and cons:
1) with DB2 SQL/XML we only have to admin, and maintain (consider
backups, security, support) only ONE datasource
2) with DB2 SQL/XML the developer's machine refer/keeps only ONE
datasource
3) with MongoDB, the querying is more simple and intuitive (sorts,
ranges, aggregations, distinct)
4) with MongoDB, we have support of SpringData(Spring Data - MongoDB | SpringSource.org) or DataNucleus (DataNucleus Access Platform - MongoDB Datastore)
to get a data source abstraction
5) with DB2 we have to create a custom type to mapping the XML
attribute, but this feature is purelly database specific (non
standard)
6) Scalability: MongoBD appears to be so scalable, and DB2 SQL/XML too
(DB2 9 XML performance characteristics)
7) HA or replication: both have replications features

Questions:
1) what other features we must consider? (besides to the above mentioned)
2) which alternative is better considering that cuestions

Any comments are welcome
Thanks you