You are viewing a plain text version of this content. The canonical link for it is here.
Posted to slide-user@jakarta.apache.org by Valeriy Podkolzin <vp...@atg.com> on 2001/05/10 02:27:00 UTC
Some performance results for xml\Test
I checked the performance of xml\Test sample and got some questionable
results with SQL data store:
1. For provided Domain.xml and Import.xml retrieving-traversing node
operation:
for (int i=0; i<nbcheck; i++) {
structure.retrieve(slideToken, "/toto/2/3/4/5");
}
gave about 17,000 ops./sec. ( all nodes preset and stored in caches).
2. Changing node uri from /toto/... to /foo/... ( creating and retrieving
new nodes inside Test instead predefined in Import.xml) gave about 1,700
ops./sec., 10 times less. Looks like Slide imposes caching only on node
uri(s) predefined in .xml .
try {
structure.retrieve(slideToken, "/foo");
} catch (StructureException e) {
System.out.println
("!!!!Objects don't exist : Creating objects ...");
subject1 = (SubjectNode) structure.retrieve(slideToken,
"/");
structure.create(slideToken, subject2, "/foo");
structure.create(slideToken, subject3, "/foo/1.txt");
........
for (int i=0; i<nbcheck; i++) {
structure.retrieve(slideToken, "/foo/2/3/4/5");
}
3. Changing node uri from /toto/... to /toto/foo/... ( like put additional
collection) gave 15,000 ops./sec.
try {
structure.retrieve(slideToken, "/toto/foo");
} catch (StructureException e) {
System.out.println
("!!!!Objects don't exist : Creating objects ...");
subject1 = (SubjectNode) structure.retrieve(slideToken,
"/");
structure.create(slideToken, subject2, "/toto/foo");
structure.create(slideToken, subject3, "/toto/foo/1.txt");
........
for (int i=0; i<nbcheck; i++) {
structure.retrieve(slideToken, "/toto/foo/2/3/4/5");
}
I didn't find the cause, but a wierd thing that in any case I saw that Slide
put a node uri ...foo... into cache in HashMapObjectCache.put(key, value)
and get the object for the key ...foo... from there for all samples.
Valeriy Podkolzin