Page tree
Skip to end of metadata
Go to start of metadata

Product versions:
Business Objects Enteprise XI 3.1

Deployment and implementation options

  • consider deploying Federation Web Services on its own Java Application server using your own web services provider so it does not interfere with other web apps like Infoview. The screenshot below gives an example on how to do that : the remote connection defined on vdcitbi02 (destination) to testafiesta3 (origin) via a separate Tomcat instance running on plbvm03.

    Note : for fine tuning and better performance, it is recommended you turn off the Federation web service where it is not required. To do so, edit the following file :

    <service name="Federator" activate="false">

  • you should be able to replicate up to 32,000 objects in each Replication Job. some deployments may need smaller or larger replication sizes. only support up to 100,000
  • smaller BIAR files with "split" options

Tuning parameters

Here is a list of things to consider when dealing with performance and heavy load, categorized by area of tuning or component involved in replication jobs (see How Federation XI 3.1 works ? for process flows and identification of main components). More recommendations around these parameters to come later this quarter.

Web Service


when should we use it ?

-Dbobj.biar.suggestSplit=<number of MB> and -Dbobj.biar.forceSplit=<number of MB>

to prevent ZipFile exception: java.lang.OutOfMemory Error

3 params in <webapps>/dswsbobje/Web-Inf/conf
<parameter name="cacheAttachments" locked="false">true</parameter>
<parameter name="attachmentDIR" locked="false">tempdirectory</parameter>
<parameter name="sizeThreshold" locked="false">4000</parameter>

File caching decreases performance as the Web Services process to files instead of memory. You may use a combination of both options and send large transfers to a file and smaller ones into memory


increase max heap size param as high as possible

"session.timeout=<value>" in <webapps>/dswsbobje/WEB-INF/classes/

When replicating a huge number of objects, we need longer web service sessions to accomodate the request

Adaptive Job Server

suggested job server command line ((warning) work in progress, this is currently under validation)
-maxChildReqs 100 -javaArgs Xmx1024m,XX:MaxPermSize=256m,XX:LoopUnrollLimit=0


when should we use it ?

-javaArgs Xmx512m

increase max heap size of each Adaptive Job Server to avoid replication job failures with OutOfMemoryError :

Unexpected exception caught. Reason: [java.lang.OutOfMemoryError: java heap space]
you can go up to 1024m. Above that, you want to add more job server instances

-javaArgs "XX:MaxPermSize=256m"

increase PermGen because of java serializable classes used by Derby that need to loaded into memory

-javaArgs XX:LoopUnrollLimit=0

to prevent a crash from job server when importing large BIAR files

-javaArgs Dbobj.federation.WSTimeout=<value>

connection timeout for JavaJobServerChild

-javaArgs Dbobj.federation.ArchiveBackupPath=<directory>

to keep a copy of each BIAR file generated during a replication job execution, for troubleshooting only

-maxChildReqs 100

By default this parameter is set to 100. It specifies the number of jobs processed by a job server child. Once the limit is reached, a new job server child process is spawned. You may want to decrease that value to avoid job failures with OutOfMemoryError.



when should we use it ?

-BatchExpirySeconds <seconds>

when committing a large number of objects, the Object Manager (within App Server) can not stream the objects quick enough to the CMS for the CMS to commit the objects within the CMS timeout window. default is 10 seconds

ThreadTimeoutAfterNMinutes in HKEY_LOCAL_MACHINE\SOFTWARE\Business Objects\Suite 12\CMS\Instances\<CMS_NAME>.cms\ConnectionTimeout in HKLM\Software\Business Objects\Suite 12\CER

When replicating a large number of objects (error occurred at approximately 300k objects, though this may vary depending on your environment), Federation will send queries to the CMS that require more than 9 minutes (default value) to be executed

  • No labels


  1. Former Member


    My customer had the issue that after 9 minutes the replication got a timeout.

    So we increased the connection timeout:86460000 which is 24 hours 1 min and
    CMSThreadTimeout: 1440 which is 24 hours.

    Still after that the replication now stopps after 104 minutes even it is set to 24 hours any ideas ?



  2. How many objects are in this replication job ?

  3. Former Member

    Apparently there are no more then 2000 objects

  4. Either those are really big or have very complex security, because 2000 is not that many and shouldn't take that long...

  5. Former Member

    I suggested to the customer to edit this :

    "session.timeout=<value>" in <webapps>/dswsbobje/WEB-INF/classes/

    he says there is not such settings

    in, i only see qaaws.session.timeout=
    is this the correct paramter? i dont see session.timeout alone? Please