[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Loading big ot.xml file in Ozone



Hi Ashvil,

Thu, 18 May 2000 Ashvil wrote:
>      I am evaluating Ozone for use for the open data directory 
>project http://i3connect.net. I plan to write a detailed set of 
>requirements soon on what we think we would need from an XML 
>database.

Looking forward to read you requirements.

>Ozone looks very interesting. I used the Ozone db 0.4 version to 
>test it. I had couple of issues with it 
>
>a. Queries took some time. I don't have exact benchmarks and 
>reading the mailing list archives, it looks like this problem has 
>been fixed by using the XT parser. 
>
>b. I tried using the Store.java program to load the Old Testament 
>XML file. It a 3+Mb file. It gave a parsed in memory message after 
>35 seconds and then threw a lot of java exceptions. The file is 
>valid XML and IE5 loads it after using 60Mb of memory space. 

Use SAXStore.class, which is in fact much more scalable than Store.class.

What is the difference?

Store.java uses a DOM Parser (Xerces or any other) to parse an XML file
and serialize the _whole_ in-memory document to the ozone server. SO you
need enough memory to serialize and deserialize this object.

SAXStore.java uses a SAX Parser (Xerces or any other) to parse an XML
file and passes the events to the ozone server. To avoid such a lot of
event passing (that produces a lot of ozone rmi calls), events are
accumulated (1000 events by default) and send as an event stack to the
ozone server.

>You can find ot.xml in this zip archive.
>http://metalab.unc.edu/pub/sun-info/standards/xml/eg/rel200.zip

Thanks for this link, it is a good testsuite besides the Shakespeare
xml testsuite.

Regards,
Lars
--
________________________________________________________________
Lars Martin                         mailto:lars@softwarebuero.de
softwarebuero m&b (SMB)              http://www.softwarebuero.de