10:42 a.m.
Hello, my name is Aleksander M. Stensby.
We are currently using Apache Lucene for our full-text indexing/search
system, but are considering a possible move over to Hibernate Search (or
possibly Compass).
I've been looking at Hibernate Search and have a question that I hope to
get some help with here (if this is the right list to ask the question,
that is.)
We currently have a pretty big and complex database system consisting of
several "similar" databases. That is, the tables and the table structure
of each of these X databses are the same, but the content is not, meaning
it is not a mirroring og sharding system.
What we do at the moment is run X different indexes, each representing a
different database. The indexing is done by batch jobs running several
times a day (we have a pretty large flow into our database every day). The
solution is not very scalable, as we have experienced, and it causes big
problems if something goes wrong with one of these batch jobs. Also, we
actually run one of these processes for each database, each dealing with
its own index. So yeah... a big mess in general.
Since we do use Hibernate, I fell in love with the idéa of Hibernate's
Event system taking care of this indexing for us, but one big and
essential question remains:
How can Hibernate "tackle" several different "identical" databases?
I mean, the layout would be something like this:
DB#1
- TableA
DB#2
- TableA
DB#X
- TableA
TableA in DB#1 would refer to entity no.integrasco.xxx.tableA, but so
would TableA in DB#2, and TableA in DB#X...
So how can we handle indexing of these entities into different indexes?
So that DB#1 - TableA goes into:
/usr/local/lucene/indexes/db1/tableA
DB#2 into
/usr/local/lucene/indexes/db2/tableA
etc..
Is it possible? Maybe I haven't spent enough time on this yet, if the
answer is simple, please forgive me for wasting your time...
Best Regards,
Aleksander M. Stensby
--
Aleksander M. Stensby
Senior Software Developer
Integrasco A/S
aleksander.stensby(a)integrasco.no
Tlf.: +47 41 22 82 72
5:26 a.m.
Hello Aleksander,
I've been in a similar situation having to decide between Compass, direct Lucene
and Hibernate Search. I ended up using Hibernate Search because we use
Hibernate and I am very satisfied now, after a year of experience.
About your specific question, I'm not sure I understood your situation.
Do you have a single entity that is saved on multiple databases? How
do you tell when saving an entity to which database it belongs to?
Are you using Hibernate Shards?
Do you want all entities indexed in the same index, or do you need to
keep the indexes separated? You could also put a "type token" in the
index, so you can filter on one of them or choose to make a "global"
search on all databases.
Have you read about Hibernate Search index sharding capabilities?
You may get better answers in the user forums:
http://forum.hibernate.org/
regards,
Sanne Grinovero
2008/2/23, Aleksander M. Stensby <aleksander.stensby(a)integrasco.no>:
Hello, my name is Aleksander M. Stensby.
We are currently using Apache Lucene for our full-text indexing/search
system, but are considering a possible move over to Hibernate Search (or
possibly Compass).
I've been looking at Hibernate Search and have a question that I hope to
get some help with here (if this is the right list to ask the question,
that is.)
We currently have a pretty big and complex database system consisting of
several "similar" databases. That is, the tables and the table structure
of each of these X databses are the same, but the content is not, meaning
it is not a mirroring og sharding system.
What we do at the moment is run X different indexes, each representing a
different database. The indexing is done by batch jobs running several
times a day (we have a pretty large flow into our database every day). The
solution is not very scalable, as we have experienced, and it causes big
problems if something goes wrong with one of these batch jobs. Also, we
actually run one of these processes for each database, each dealing with
its own index. So yeah... a big mess in general.
Since we do use Hibernate, I fell in love with the idéa of Hibernate's
Event system taking care of this indexing for us, but one big and
essential question remains:
How can Hibernate "tackle" several different "identical" databases?
I mean, the layout would be something like this:
DB#1
- TableA
DB#2
- TableA
DB#X
- TableA
TableA in DB#1 would refer to entity no.integrasco.xxx.tableA, but so
would TableA in DB#2, and TableA in DB#X...
So how can we handle indexing of these entities into different indexes?
So that DB#1 - TableA goes into:
/usr/local/lucene/indexes/db1/tableA
DB#2 into
/usr/local/lucene/indexes/db2/tableA
etc..
Is it possible? Maybe I haven't spent enough time on this yet, if the
answer is simple, please forgive me for wasting your time...
Best Regards,
Aleksander M. Stensby
--
Aleksander M. Stensby
Senior Software Developer
Integrasco A/S
aleksander.stensby(a)integrasco.no
Tlf.: +47 41 22 82 72
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
3:19 p.m.
The right place for those questions is http://forum.hibernate.org
To answer your question, you have different DBs, so different
SessionFactory configurations (using the same mapping I imagine).
You will just have to change the following property in each
hibernate.cfg.xml
hibernate.search.default.indexBase /usr/local/lucene/indexes/dbXX
Please follow up on the forum if you have further questions.
On Feb 23, 2008, at 11:42, Aleksander M. Stensby wrote:
Hello, my name is Aleksander M. Stensby.
We are currently using Apache Lucene for our full-text indexing/
search system, but are considering a possible move over to
Hibernate Search (or possibly Compass).
I've been looking at Hibernate Search and have a question that I
hope to get some help with here (if this is the right list to ask
the question, that is.)
We currently have a pretty big and complex database system
consisting of several "similar" databases. That is, the tables and
the table structure of each of these X databses are the same, but
the content is not, meaning it is not a mirroring og sharding system.
What we do at the moment is run X different indexes, each
representing a different database. The indexing is done by batch
jobs running several times a day (we have a pretty large flow into
our database every day). The solution is not very scalable, as we
have experienced, and it causes big problems if something goes
wrong with one of these batch jobs. Also, we actually run one of
these processes for each database, each dealing with its own index.
So yeah... a big mess in general.
Since we do use Hibernate, I fell in love with the idéa of
Hibernate's Event system taking care of this indexing for us, but
one big and essential question remains:
How can Hibernate "tackle" several different "identical" databases?
I mean, the layout would be something like this:
DB#1
- TableA
DB#2
- TableA
DB#X
- TableA
TableA in DB#1 would refer to entity no.integrasco.xxx.tableA, but
so would TableA in DB#2, and TableA in DB#X...
So how can we handle indexing of these entities into different
indexes?
So that DB#1 - TableA goes into:
/usr/local/lucene/indexes/db1/tableA
DB#2 into
/usr/local/lucene/indexes/db2/tableA
etc..
Is it possible? Maybe I haven't spent enough time on this yet, if
the answer is simple, please forgive me for wasting your time...
Best Regards,
Aleksander M. Stensby
--
Aleksander M. Stensby
Senior Software Developer
Integrasco A/S
aleksander.stensby(a)integrasco.no
Tlf.: +47 41 22 82 72
_______________________________________________
hibernate-dev mailing list
hibernate-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hibernate-dev
6466
days inactive
6467
days old
2 comments
3 participants
participants (3)
-
Aleksander M. Stensby -
Emmanuel Bernard -
Sanne Grinovero