Does anyone know if it is possible to modify the hibernate mapping during runtime?

I am using Hibernate 3, and here is a very brief rundown of what I want to accomplish.

I have a system which has a set of 4 tables. Each client on the system has their own table set which are all suffixed with their client ID (tbl_myTable_000000 or tbl_myTable_111111).

My Java app is going to have a Thread per Client model. Where each thread will be accessing its tbl_myTable_[client id] tables for processing. There is no reason I need to have a set of Java classes for each client, because they are all going to be the same. I know I will need a SessionFactory per Thread, but what I would like to do is have it load the same base configuration files, then modify them at runtime to map to the correct _[client id] tables.

Is the modification/reload of the configuration possible at runtime?

Thanks in advance!
Geoff

anyone?

Just want to say why do you create different tables and why don t you have a clientid on your tables two know to which client the record belong.

Maybe runtime manipulation work (you can try it) but it does not seem to me to be a good design choice

Seb

Additionally you can use a session filter to restrict all queries to one client id.

See http://www.hibernate.org/hib_docs/v3/reference/en/html/filters.html#objectstate-filters

I have been designing databases for a very long time... and have decided that this is the way to go.

I have a couple of good reasons why I have decided to create client specific tables as opposed to using key relationships...

1) The tables will be re-indexed, truncated, etc... things that would temporarily lock a table.

2) Each client will have millions of records. I would like to keep it as simple as possible to move a client to a new physical piece of hardware if they ever beging using too much processing power.

3) There are going to be many columns that aren't shared between clients... but these will not be "known" by my Java application running Hibernate... so the model still works.

I am very familiar with database normalization... so that isn't the point here. I am just wondering about runtime modification of the mapping files...

Thanks,
Geoff

Ok, this is just a question.

If the number of client is known at compile time you can also create a bunch of hbm files from a core one with a ant script if runtime modification isn't possible (I don't know if it is)

I would prefer to not have to rebuild the code each time I add a new client. The system will peridocally check which clients exist, and automatically configure itself to begin or terminate servicing those clients in near real time. I am digging through the many layers of hibernate right now looking for something that looks like it would allow me to modify mapping documents at runtime... no luck so far :(

Ok.. well I have figured it out... it took a bit of hacking, but I got it.


/*
* *************************************************
* *************************************************
*/
import org.hibernate.mapping.*;
import org.hibernate.cfg.*;
import org.hibernate.*;

...
...
...

Configuration configuration = new Configuration();
Iterator i = configuration.getTableMappings();

while(i.hasNext()){
Table table = (Table)i.next();
table.setName(table.getName() + "_000000");
}

SessionFactory factory = configuration.buildSessionFactory();
Session sess = factory.openSession();

/*
* *************************************************
* *************************************************
*/


That should do it... now the SessionFactory will automatically be pointing to the _0000000 tables for the application. Probably what I will do is load the table set into a map and update them based on the need... seeing as there are some shared tables in my schema that don't have the _clientID.

Thanks for everyones help.

geoff