From: that.SelectedDude <tha...@gm...> - 2006-09-27 19:24:54
|
Hi there, and thanks for this marvellous product, and for any support. - This is an absolute beginner's question, so please be patient. I've downloaded eXist 1.1, created a new collection, and uploaded an XML file (BMEcat catalog, 20 MB (including whitespace), 7000 articles) into that. Using the sandbox application, I've queried for a) //ARTICLE b) //ARTICLE[SUPPLIER_AID = '211180'] The first query is lightning fast, whereas the second one takes longer (0.3, resp., 1.2 seconds, apparently depending on CPU speed (tested on two computers)). This for one is surprising - as SQL statements, in contrast, usually have their bottlenecks in the amount of output data (and, of course, in the number of JOINs involved). OK, here, it apparently is, how specific the query is. - I was just wondering... - However, my actual question is about indexing. - I've used the Java client to create a new range index, //ARTICLE/SUPPLIER_AID When I closed the form, two confirm boxes popped up: "Save changes?" - "yes" "Reindex?" - "yes" Next, nothing happened (no CPU or harddisk usage, no improvement in the speed of the queries). OK, next, I used the Java client's "File --> Reindex collection" tool, and there *was* CPU and harddisk usage for a while. But back in the sandbox application, the query //ARTICLE[SUPPLIER_AID = '211180'] now takes forever (doesn't return within a larger number of minutes)! (The query //ARTICLE however is still fast.) Could anybody give me a clue, please? Thanks indeed. Cheers Robert ---------- The environment, BTW: Ubuntu Linux, Sun Java 1.5 P.S.: I've looked a the install folders but haven't been able to find the data storage files, the index storage files, or any definition files... Thanks again. |