Last updated on APRIL 10, 2017
Applies to:Oracle Tuxedo - Version 12.1.3 and later
Information in this document applies to any platform.
Weblogic calls Tuxedo using a Tuxedo JOLT call to fetch a large file.
Currently, Weblogic makes multiple calls to a service in Tuxedo to fetch a file in batches of approximately 10k bytes.
Attempting to improve performance the buffer size was changed to 100k and 200k but this resulted in degradation of performance.
The transaction which takes an avg of 8 secs, with a 10k byte buffer (using multiple JOLT calls), is taking 40 secs and 60 secs with 100k and 200k buffer respectively.
After detailed investigation it was found that the maximum time is spent on Weblogic side while decoding the Tuxedo service output response.
From thread dumps the following stack appears to be the cause of the latency:
tid=0x037eb0b8 nid=0x7c runnable [0xa48dd000]
- locked <0xb81052c8> (a java.util.Vector)
- locked <0xb8110b00> (a bea.jolt.JoltMessage)
Size of response message sent back by the service in Tuxedo is greatly increased.
Sign In with your My Oracle Support account
Don't have a My Oracle Support account? Click to get started
My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms