Hi Thanks, this little file contains 300.000 records. So your input file must be something like 433 GB I assume that you loaded the file with LOCKED mode, then at least you have a chance without the temporary copies needed. With 9 fields of 8 bytes each, your storage footprint becomes easily 8 * 133 GB = 1.2TB, excluding the string value dictionaries. Strings may use less then 8 bytes each when there are a lot of duplicates. To be on the safe size, I would aim for free disk-space of about 3-4TB You might consider having a look at http://www.monetdb.org/Documentation/Cookbooks/SQLrecipes/storage-model to get a better estimate after loading your small sample. On 8/28/13 10:48 PM, Ralph Holz wrote:
Hi,
I have made a very short input file available on
http://pkidata.net.in.tum.de/export.csv.tgz (33M) The delimiter is #.
Looking at that file again, I must have missed some fields in the schema in my last mail. I think you can ignore those, they're mostly empty. COPY into does not ignore them, they have to be specified. There is not (yet) a partial COPY into provided.
I have no clue what will happen if your scheme does not match the file, at least a good test. regards, Martin
Ralph