[xquery-talk] performance figures for xquery processors

Jonathan Robie jonathan.robie at datadirect.com
Tue Feb 7 16:35:29 PST 2006

Ronald Bourret wrote:

> For a list of other benchmarks, see:
>    http://www.rpbourret.com/xml/XMLDBLinks.htm#Benchmarks
> But like all benchmarks, you should take them with a grain of salt. 
> What's important to you is whether or not a processor can perform your 
> queries fast enough for your needs.
> I also agree with Martin's statement that you should look at native 
> XML databases. If you are running XQuery queries on documents received 
> at run time, such as routing documents based on the result of a query, 
> they won't help. But if you are querying documents stored over time, 
> the pre-parsing and indexing they provide will significantly improve 
> your query performance.

Here are some distinct usage scenarios that probably deserve separate 

1. Creating XML structures from relational data. Like it or not, this is 
a very common scenario, partly because relational databases are so 
widely deployed, and partly because recombining data to create different 
structures is often very important. XMark and several other of the 
benchmarks Ron points to seem to address this scenario (though better 
benchmarks could be designed).

2. Returning entire XML documents, Extracting portions of very large XML 
documents. (which many XML databases do very well)

3. Joins between nodes of large XML documents.

4. Returning large or complex XML structures, either by direct 
extraction (as in 2) or as the result of more complex operations (as in 
1 or 3).

5. Full-text queries (which MarkLogic will do particularly well)

What else do y'all think should be addressed by standard benchmarks?


Read my Blog: http://blogs.datadirect.com/jonathan_robie/
Learn XQuery: http://media.datadirect.com/download/docs/ddxquery/tutorial_query.html
Learn XQJ (the JDBC for XQuery): http://www.datadirect.com/developer/xquery/topics/xqj_tutorial/
Get DataDirect XQuery: http://www.datadirect.com/products/xquery/

More information about the talk mailing list