From: Romanowski, Tim <tim.romanowski_at_lmco.com>
Date: Sat, 28 Apr 2007 19:42:59 -0400

I've been reading up Persistence Units and deployment, and have a few
questions on the following scenario:

 

Server A hosts application MyWebApp using Glassfish. My WebApp uses the JPA
to persist its data to its own local database.

Server B also hosts MyWebApp using Glassfish. Similarly, it persists its
data to its own local database.

 

Server B is rarely used, and does not hold that much data. In fact, it
isn't even hooked up to the network very often. On the rare occasion that
Server B is used, I would like to merge all OR some of the data from Server
B's database back into Server A. The trick here, is that I would like to
provide an 'automated' way to merge this data, so that an administrator on
Server B can go to a webpage in the webapp, select which tables to transfer,
click a link, and have the data saved to Server A's database.

 

1) I could use a web service to transfer the data, but that seems like it
would be very inefficient.

2) Another option is to have MyWebApp on server B just call a script and
dump the data to a file. But that means manually (or via script) merging
that file back into the DB. I don't want to do that.

3) Lastly, what I really want to do (or is this bad?) is have Server B
connect to Server A's datasource (keep in mine, both web apps have the same
entity classes), and then use the POJO love from the JPA to get all entities
from Server B and insert them into Server A's database. It wouldn't be a
direct copy in the since I'd obviously be using two different EntityManagers
to access a given table in both databases, but I could (worst case) insert
each entity from Server B into Server A.

 

Is there a slick way to implement number 3? My biggest problem is that
neither server knows the OTHER server's database information (such as ip
address) until runtime. From what I've read, I can't tell if it's possible
to define a persistence unit's details at runtime, and even less likely, I
wonder what the issue would be with trying to uses the same entity classes
in a single web app (on a single server) to connect to two separate PUs.
The idea is simple, have one app that connects to two identical databases,
instantiate entities from one database and save them into another. The
reason I want to do this automagically in the web app, is so I can perform
some business logic on Server B's data before it is inserted into Server A.
Is this a flawed approach? Any suggestions?

 

 

Tim