Speed up OpenSAML initialization ?

[Brent Putman]

On 2/27/15 2:43 PM, Cantor, Scott wrote:
> On 2/27/15, 2:38 PM, "Brent Putman" <putmanb at georgetown.edu> wrote:
>> That seems like a long time.  I've never noticed that it takes that long 
>> for me, but I'll pay attention the next time I'm running tests or 
>> starting my test IdP.
> Me either, though of course the tests spinning up can take time. But that 
> seems more likely to be related to the XML provider initialization and 
> such.

Yeah, I just ran a test with logging timestamps turned on.  The
AlgorithmRegistry init took 120ms.  And this is on my workstation in
Eclipse.  So I'm very skeptical that the time is there.

I think Scott is right, it's much more likely to be in something like
the XML provider init.  But I still have never seen it take anywhere
near that long.

Below is a run on my workstation from a test in opensaml-saml-impl with
logging timestamps. Took a little over 3 seconds (not counting the
Test* ones at the end which wouldn't run normally). Seems like most of
the time is taken by the 3 big XMLObject provider initializers
(highlighted).  The SAML one for example took about 1.5 seconds, The
xmlsec and SOAP provider ones are similar. Those are the ones that have
lots of XMLObject provider impls. I don't know what we can do about those.

The AlgorithmRegistry took about 212ms here. It's not a problem.



2015-02-27 15:14:11,728 --  INFO  InitializationService - Initializing
OpenSAML using the Java Services API
*2015-02-27 15:14:11,784 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.saml.config.XMLObjectProviderInitializer*
2015-02-27 15:14:13,418 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.saml.config.SAMLConfigurationInitializer
2015-02-27 15:14:13,418 --  DEBUG SAMLConfigurationInitializer -
Initializing SAML Artifact builder factories
2015-02-27 15:14:13,436 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.core.xml.config.XMLObjectProviderInitializer
2015-02-27 15:14:13,466 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.core.xml.config.GlobalParserPoolInitializer
2015-02-27 15:14:13,469 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.xmlsec.config.GlobalAlgorithmRegistryInitializer
2015-02-27 15:14:13,681 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.security.config.ClientTLSValidationConfiguratonInitializer
2015-02-27 15:14:13,844 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.xmlsec.config.JavaCryptoValidationInitializer
*2015-02-27 15:14:13,845 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.xmlsec.config.XMLObjectProviderInitializer*
2015-02-27 15:14:14,239 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.xmlsec.config.ApacheXMLSecurityInitializer
2015-02-27 15:14:14,240 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.xmlsec.config.GlobalSecurityConfigurationInitializer
*2015-02-27 15:14:14,346 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.soap.config.XMLObjectProviderInitializer*
2015-02-27 15:14:15,150 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.core.xml.config.TestXMLObjectProviderInitializer
2015-02-27 15:14:15,221 --  DEBUG InitializationService - Initializing
module initializer implementation:
org.opensaml.xmlsec.config.TestXMLObjectProviderInitializer

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://shibboleth.net/pipermail/dev/attachments/20150227/ed0595e3/attachment.html