I want to obtain the number of XML objects in one doctype. But it takes a long time for Tamino to count all the objects (for a million of objects).
So I think if I can create index for ino:id’s then I probably increase the performance when processing all the XML objects in one doctype.
Does anybody know whether it it available in Tamino to create index for ino:id’s or somehow to increase performance when processing all the XML objects in sequence?
No, it is not possible to index ino:id - it makes no sense to. Each document that is stored is automatically associated with a unique document number - the ino:id. This is used to directly access the document, so a query like a[@ino:id=12] should be the fastest access method to a document.
The one thing that you have not mentioned is what version of Tamino you are using. Tamino v2.3 has performance enhancements so maybe upgrading to this version will help quite alot.
I am using Tamino 18.104.22.168.
There are now 1 million XML objects in my doctype.
I ran such a query through IE 5.0 on doctype book of collection bib on Windows 2000:
It took 5 minutes and 14 seconds to inosrv.exe to perform this task. (Elapsed time is longer surely)
This is the case.
The current version 22.214.171.124 on Windows should reply very fast on this query. There was a performance fix for count() on root level.