Shahbaz wrote:
I was finally able to install monetdb and put data into it (on OSX), although I don't have any performance numbers to report back just yet. Thanks. Good to know
-If i delete or even drop a large table from MDB5, many of the BAT files remain on the file system. Is there a command that deflates/shrinks or completely removes to files? I wasn't able to find anything in the docs
This sounds not so good and should not happen. It would be nice if you can demonstrate this with a small example and file a bug report. If you drop a table the underlying files should have been gone. If you delete, the files remain but will be empty. [Aside, we noticed some memory leakage, which has been attacked recently. This has been resolved in the bug-fix release emitted 2 days ago] Suggested quick approach. Dump your SQL database , throw away dbfarm and reload the SQL database.
-According to previous messages on the mailing list and some other docs, MonetDB 5 is supposed to work with distributed databases (Armada?). I don't see anything regarding this in the docs either, is it coming soon?
Armada is a research project on distributed and autonomous DBMS. There won't be a releasable version soon. Currently, distribution would have to be taken are of in the application/middleware. We haven't tried out JDBC-based stuff like GORDA http://gorda.di.uminho.pt/publications
-I am very (VERY) interested in the ability to have a cluster/grid of MDB servers which let me distribute data and computation while providing an easy to use interface for end users and admins. Along
Yes we too and are working on innovations in this area in the context of scaling up the skyserver implementation(http://cas.sdss.org/dr5/en/)
with that, if we get the ability to process streaming data (such as stock quotes), MonetDB will become king of Wall Street :) yes is looked into with 2 PhDs.
-I see that one of the recommended projects for students is to extend MonetDB to process time series data. If someone is looking at this, please take a look at this paper: http://citeseer.ist.psu.edu/610025.html (AQuery for ordered data). Yes, array based support is part of our research agenda for some time for information retrieval. The latest paper available from the website: R. Cornacchia, S. Heman, M. Zukowski, A. P. de Vries, P. A. Boncz. Flexible and efficient IR using array databases. Technical Report INS-E0701, CWI, Amsterdam, The Netherlands, January 2007.
regards, Martin