Not that I know of, and I am going to be the annoying person who tells you to 
rethink that "dedicated forest" approach.

The design of MarkLogic is to place documents into forests automatically. This 
creates balanced, fairly homogenous forests. If you try to manage document 
forest assignment manually, you will probably create unbalanced, heterogeneous 
forests. You may find yourself fighting with the software in other places, too.

Sometimes developers new to MarkLogic think that it would be a good idea to put 
documents of type A in forest A, and type B in forest B: to dedicate forests by 
content. Even http://docs.marklogic.com/guide/ingestion/xquery mentions some 
"advantages" to this approach. But I think it is misguided, and nearly always a 
mistake. Before embarking down that path, you should understand the potential 
disadvantages of that approach.

Consider what happens when you query heterogeneous forests. If you have a lot 
of queries on documents of type A, forest B is relatively idle. Because forests 
operate in parallel, this is a net loss vs concurrency across multiple forests. 
So this approach throws away much of the advantage of multiple CPU cores, and 
multiple disk subsystems too. This approach also makes query development more 
awkward, because querying by forest is not as straightforward as querying by 
collection or directory.

IN my opinion the use-cases mentioned at 
http://docs.marklogic.com/guide/ingestion/xquery are better addressed using 
other product features. If you only want to back up the latest updates, use 
journal archiving. If you want to mix fast storage and slow storage, set up the 
faster storage as a fast data directory.

So I prefer to let the system place documents wherever it likes, and use 
directories or collections to manage them. When each database forest contains 
documents of different types, every query will parallelize across the database 
forests.

-- Mike

On 28 Jul 2013, at 22:46 , sini narayanan <[email protected]> wrote:

> Hi All,
> 
> Using MarkLogic content pump, is there any option to load documents into 
> dedicated forest of a database?
> 
> Thanks,
> Sini
> _______________________________________________
> General mailing list
> [email protected]
> http://developer.marklogic.com/mailman/listinfo/general

_______________________________________________
General mailing list
[email protected]
http://developer.marklogic.com/mailman/listinfo/general

Reply via email to