Tamino XML server performance

, ,
1

While working with Tamino I have encountered several problems.
I’m using Tamino version 2.3.1.1 for windows 2000,hardware:
Intel p3 500mhz cache:512kb main memory:1024mb
Tamino database size: small- about 20,000 XML and Non XML(images)documents stored.
Data base Properties are set to default.

1.The cpu resources needed in order to locate a single document are very high, a query brings the cpu usage up to 100% ,
query result is accepted only after few seconds.
This also happens when using the count function within query.
Indexes are well defined(search defined as standard).
Is there a way to better results for these queries.


2.I’m using the Java API in order to load documents to Tamino, about 20,000 documents and images(to a different collection)
are loaded every time.
Session openes(and closes after loading or update) for each document,image or update.
for every document loaded update is being performed in another collection.
After few thousands documents, loading stops.
After this, querying the collection using the interactive interface gives this
result message:“ino:messagetext ino:code=“INOXYE9499”>
Transaction aborted due to non-activity timeout or administration intervention.”
(non-activity timeout is set to 900 secondes.)
This message apears with every query,and the only solution is to restart Tamino
data base.
What might be causing this problem?

regards,
Guy Sh

2

It is quite impossible to answer question 1)
without an example schema and a query that runs slowly - an representative instance would also be helpful.

A couple of general points - might be useful

One possible cause is that, though the schema is correctly indexed, the query supplied can’t take advantage of the indexing

suppose the value of element a is standard indexed then the query

mydoctype[a~=“book”] will be slow - because the node is not text indexed.

similarly

//a[.=“book”]

may be very slow because even though “a” is indexed correctly in your target doctype - Tamino will search other doctype for “a” elements and will even search EACH instance in your doctype for an “a” elements in positions that are not indicated in the schema - unless it thinks the schema is closed. If a schema is open then Tamino admits the possibilty of elements which are not in the schema.

In Tamino 3.1 - the //a problem is addressed with the structure index and with open and closed schemas.