John we use faily large xml schema and when we started caching class descriptor and unmarshalling time went down from 215ms to 25ms average. Also when I replaced excerces parser with Piccolo we saved another 5-7 ms. Ramesh
John Weir wrote: > This is Just Info that I thought others might be interested in .. > > I was interested to know how much improvement caching the descriptor classes > made. I tested it on a reasonably large set of files each wit over 500 types > and 40% of those repeating. > > You gain between 55% and 75% benefit by Caching the descriptors. If anyone > has similar anecdotes surrounding performance, I for one would be > interested. > > Thanks, John > > > -----Original Message----- > > From: Keith Visco [mailto:[EMAIL PROTECTED] > > Subject: Re: [castor-dev] System freezes when uploading bigger XML file > > using Castor > > > > > > > > In addition to the previous suggestions by Erik and Milan, > > > > To improve performance in general, you can also try the following: > > > > 1. Disable validation > > > > unmarshaller.setValidation(false); > > > > 2. Cache the ClassDescriptorResolver (which holds onto the in-memory > > descriptors/mappings) > > > > import org.exolab.castor.xml.ClassDescriptorResolver; > > import org.exolab.castor.xml.util.ClassDescriptorResolver; > > > > ClassDescriptorResolver cdr = new ClassDescriptorResolverImpl(); > > ... > > > > while (...more stuff to unmarshal...) { > > Unmarshaller unmarshaller = new Unmarshaller(...); > > unmarshaller.setResolver(cdr); > > } > > > > 3. If you're using a mapping file, make sure it's loaded only once. > > > > 4. To reduce object creation/deletion you might want to try out the > > *experimental* object-reuse feature, by reusing your existing > > in-memory object > > model from the last unmarshal: > > > > Unmarshaller unmarshaller = new Unmarshaller(myObject); > > unmarshaller.setReuseObjects(true); > > > > Be careful though, this feature may have problems with complex > > object models (it's an experimental feature). > > > > 5. Try using different XML parsers, some are faster and more efficient > > than others. > > > > --Keith > > > > > > > > "Beg, Meraj" wrote: > > > > > > Hi, > > > Using castor, I am try to upload multiple XML files into Oracle > > database. > > > Once Java Objects are created in the memory, I use Toplink to > > actually write > > > these objects into the database. > > > I am running it as a Java Application on DOS prompt and passing > > folder name, > > > where all my XML files are residing, as a parameter. > > > > > > Problem I am facing here is, when any of the XML file exceeds > > 20,000 records > > > of average length, it hogs/Freezes whole production system. > > > > > > My Production system has following configuration for time being: > > > Windows 2000, IBM WebSphere 4.0 with 2 GB of RAM. > > > > > > anyone, please help me in optimizing castor to perform this task > > > successfully. > > > > > > Or atleast, let me know some guidelines to optimize Castor performance. > > > > > > Thanks, > > > Meraj > > > > > > ----------------------------------------------------------- > > > If you wish to unsubscribe from this mailing, send mail to > > > [EMAIL PROTECTED] with a subject of: > > > unsubscribe castor-dev > > > > ----------------------------------------------------------- > > If you wish to unsubscribe from this mailing, send mail to > > [EMAIL PROTECTED] with a subject of: > > unsubscribe castor-dev > > > > > > ----------------------------------------------------------- > If you wish to unsubscribe from this mailing, send mail to > [EMAIL PROTECTED] with a subject of: > unsubscribe castor-dev ----------------------------------------------------------- If you wish to unsubscribe from this mailing, send mail to [EMAIL PROTECTED] with a subject of: unsubscribe castor-dev
