Hi Frederic,

On 10/05/2013 14:50, Frederic Toublanc wrote:
Looking at your code I notice you start a transaction near the start of saveElement and then have another begin/commit on each nested call to updateProperty. Is that what you meant to do?

No sorry i forgot to remove the updateProperty part.

What is strange is that when i juste create the TDB with this code, it's size is 192MB and not 192KB ...
That is odd.

I'm pretty confident that an empty TDB database does not occupy 192MB.

Have you tried deleting the TDB directory to make sure the old TDB is gone before running your program? Also fixed the transaction issue.

I can't see the image you sent. I'm trying to find a way to confirm the size on disk. If you have deleted and recreated the TDB directory how about doing a DIR or ls -l in a command prompt to see what files are there and how big they are. On my machine there is a bunch of 8kb files. If you see something different, copy/paste the text into a reply.

If that confirms you have some huge files in there, then this problem is likely beyond my expertise.

Brian




File directory = new File(databasePath);
if (!directory.exists()) {
directory.mkdir();
}

ARQ.setExecutionLogging(InfoLevel.NONE);
ARQ.setStrictMode();

// Direct way: Make a TDB-back Jena model in the named directory.
ds = TDBFactory.createDataset(databasePath);


2013/5/10 Frederic Toublanc <frederic.toubl...@telemis.com <mailto:frederic.toubl...@telemis.com>>

    OK thats very strange because my empty TDB is not 192kb but 192MB
    as u can see in the screenshot below ...

    Images intégrées 2


    2013/5/10 Brian McBride <br...@epimorphics.com
    <mailto:br...@epimorphics.com>>

        Hi Frederic,


        > sorry MegaOctet = MegaBytes (in french)

        Merci.  Pas probleme.

        On my machine, Windows reports an empty TDB directory as
        occupying 192KB.

        I have loaded 10 variations of the RDF you provided into TDB
        using tdbloader.  At the end of this operation Windows reports
        the tdb directory size to be 198 kb.  TDB contains 212
        triples.  If I load 20 variations I get 422 triples and a size
        of 227KB.

        That is a lot less the MB you are reporting.

        Looking at your code I notice you start a transaction near the
        start of saveElement  and then have another begin/commit on
        each nested call to updateProperty.   Is that what you meant
        to do?

        The documentation for write transactions [1] says

        [[
        Once |dataset.commit()| or |dataset.abort()| is called, the
        application needs to start a new transaction to perform
        further operations on the dataset.
        ]]

        You have at least read operations outside a transaction.  I
        don't know enough about TDB to know whether that might cause
        problems.

        Your example isn't complete so I can't easily run it myself.

        Brian


        [1]
        
http://jena.apache.org/documentation/tdb/tdb_transactions.html#api-for-transactions





        On 10/05/2013 12:51, Frederic Toublanc wrote:

            Sorry my message was too long and couldn't be sent.
            Here is the file with all the information : rdf and
            implementation methods.




            2013/5/10 Bill Roberts <b...@swirrl.com
            <mailto:b...@swirrl.com> <mailto:b...@swirrl.com
            <mailto:b...@swirrl.com>>>


                (Brian: I think the usual French word for 'byte' is
            'octet', hence
                I'm assuming mo = megabyte.  Frederic, is that correct?)

                Frederic, my experience is that an empty TDB directory is
                typically about 200MB because of the default space
            allocated to
                the various indexes.

                An example with 40 million triples is about 24GB, so
            an average of
                600 bytes of total storage per triple.

                Best regards

                Bill




                On 10 May 2013, at 11:18, Brian McBride
            <br...@epimorphics.com <mailto:br...@epimorphics.com>
                <mailto:br...@epimorphics.com
            <mailto:br...@epimorphics.com>>> wrote:

                > Hi Frederic,
                >
                > On 10/05/2013 09:52, Frederic Toublanc wrote:
                >> Hello everyone,
                >>
                >> I really need some info about the size of the jena TDB.
                >> We need to be sure that this triplestore will fit
            to our needs,
                can someone
                >> please answer the folowwing question :
                >>
                >> The size of our jena TDB is increasing as hell and
            we only
                insert light
                >> data (2kb per element when written to rdf).
                >> At the begining the TDB is 192 mo and after
            inserting 15 of
                those elements
                >> it has grown to 216 mo.
                >> Is that normal ?
                > I've used TDB a bit and not had a problem with
            execessive size.
                >
                > What's an "mo"?
                >
                > If you have a minimal example, I could try to reproduce.
                >
                > What data are you adding?
                > How are you adding it?
                > Are you starting from an empty TDB?
                > What OS are you running on?
                >
                > Brian
                >
                >>
                >> Thx in advance.
                >>
                >
                >
                > --
                > Epimorphics Ltd (http://www.epimorphics.com)
                >
                > Epimorphics Ltd. is a limited company registered in
            England
                (number 7016688)
                > Registered address: Court Lodge, 105 High Street,
            Portishead,
                Bristol BS20 6PT, UK
                >




-- Epimorphics Ltd (http://www.epimorphics.com)

        Epimorphics Ltd. is a limited company registered in England
        (number 7016688)
        Registered address: Court Lodge, 105 High Street, Portishead,
        Bristol BS20 6PT, UK





--
Epimorphics Ltd (http://www.epimorphics.com)

Epimorphics Ltd. is a limited company registered in England (number 7016688)
Registered address: Court Lodge, 105 High Street, Portishead, Bristol BS20 6PT, 
UK

Reply via email to