Hi,
i'm using Sedna DB XML, i've installed and configured the application in a
CentOS system on a phisical machine with 32GB RAM and 1 TB of Hard-disk.
My actual DB-Collection contains 130Gb of XML and the Sedna configuration
is:
<?xml version="1.0" standalone="yes"?>
<db>
<name>cerved_test</name>
<bufs_num>32000</bufs_num>
<max_log_files>60</max_log_files>
<tmp_file_initial_size>6000</tmp_file_initial_size>
<upd_crt>0.25</upd_crt>
</db>
This is a part of XML i want to index (xml is contained in a Collection):
<?xml version="1.0" encoding="ISO-8859-1"?>
<s2xResponse>
<s2xStatus Reload="N">
<s2xData>[XML-NODES]</s2xData>
<s2xInfo>[XML-NODES]</s2xInfo>
<s2xDocument>[XML-NODES]</s2xDocument>
<IfisInfo>
<insert_date>20131023</insert_date>
<ndg>231459</ndg>
<group_id>231459</group_id>
<fiscal_code>07530530018</fiscal_code>
<document_date>-</document_date>
<hash_key>-</hash_key>
<is_last>true</is_last>
</IfisInfo>
</s2xResponse>
I've tried to create some indexs on dark-green keys to increase
performance in query execution:
CREATE INDEX "bs_index_ndg"
ON collection("CERVED_FULL")/s2xResponse//IfisInfo BY ndg
AS xs:string
USING "bstrie" &
CREATE INDEX "bs_index_fc"
ON collection("CERVED_FULL")/s2xResponse//IfisInfo BY fiscal_code
AS xs:string
USING "bstrie" &
CREATE INDEX "bs_index_last"
ON collection("CERVED_FULL")/s2xResponse//IfisInfo BY is_last
AS xs:string
USING "bstrie" &
I'm using a heavy query to test performance:
for $a in collection('CERVED_FULL')/s2xResponse
where $a/IfisInfo/is_last/text() = "true" and (
$a/IfisInfo/ndg/text() = "value1" or
$a/IfisInfo/ndg/text() = "value2" or
...
...
...
[~4000 OR value conditions]
)
return
<Response>
<Key1></Key1>
<Key2></Key2>
<Key3></Key3>
</Response>
This query give me result in ~240 seconds, the same time before and after
index creation.
Is something wrong in my INDEX creation?
Is there any configuration parameter i can change to increase performance?
Every day i load ~100 new xml document on db and the same query, with the
same number of OR condition (identical query), lose some time in execution,
so i think is an INDEX problem.
Any suggestions?
Thank you very much
Matteo
------------------------------------------------------------------------------
November Webinars for C, C++, Fortran Developers
Accelerate application performance with scalable programming models. Explore
techniques for threading, error checking, porting, and tuning. Get the most
from the latest Intel processors and coprocessors. See abstracts and register
http://pubads.g.doubleclick.net/gampad/clk?id=60136231&iu=/4140/ostg.clktrk
_______________________________________________
Sedna-discussion mailing list
Sedna-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/sedna-discussion
