Converting a standalone SOA/AIA instance to a cluster instance.

Hi Architects,
We have a SOA/AIA11g instance running as a standalone instance with DR configured. Now team wants to convert this standalone instance to a 2 node clustered instance.
Can someone let us know if it is possible to convert a standalone SOA instance to a clustered instance.
Thanks,
Edited by: user1144870 on Jun 14, 2011 4:44 AM

That is quite some work.
You can add a cluster by using the console and add the soa_server to the cluster.
The problem is that when you add another server to the cluster the resources (such as JMS) are targetted to the soa_server
are not targetted to the other server in the cluster. So you have to retarget all the resources to the cluster.
The trouble you are going to run into are the distributed JMS resources (such as queues and topics) for those
you have to add new resources and target them to the right server instance.
Some extra information can be found here: http://middlewaremagic.com/weblogic/?p=6872
I think the best way to proceed is the configure a new clustered environment an example of the configuration steps
involved can be found here: http://middlewaremagic.com/weblogic/?p=6040

Similar Messages

How do I deploy dependent jars in a standalone soa server?

Hi,
I have deployed a webservice on a standalone SOA server. When I try to execute it I see the following message:
faultstring>JBO-25002: Definition oracle.apps.cdm.tcaModel.parties.core.model.publicEntity.PartyPEO of type Entity Definition not found</faultstring>
Dev is saying that we need the dependent jar files on the soa server and has given me some jar files? When I put the jar files in the applib folder, I am not able to restart the webservice.
Can someone help me on how to resolve this with some detailed steps or pointers? Is that the correct location for putting the jar files? Appreciate your help.

Hi,
to me the problem appears that you didn't add the class to the WebService as an additional class to load with the Web Service. At least i JAX-RPC services you can tell the WS configuration which additonal classes are needed. In your case the PartyPEO isn't available (make sure its serializable)
Frank

Parent Instance creating many child instances in SOA-INFRA

Hi,
We have deleted a parent instance from SOA-INFRA database.Even after deletion it keeps on creating child instances. How to identify all the child instances of the parent instance in SOA-INFRA schema . Also, we searched a lot but were unable to get much details about SOA_INFRA schema.
Please suggest and tell us the reasoning behind why many child instances getting created? Any suggestions will help us.
Thanks in advance.

It just provides the table names but doesnt provide information like which table is referred which table. Also, can anyone let me know how the instance id(when a request is submitted) is created in SOA-INFRA schema and how it is used other tables because we still see the child instances is getting created from the parent instance.
Any help will be appreciated.
Thanks.

Convert ASM single instance to RAC ASM instance...

Hi, all
Im installing OracleDatabase with RAC and I got an error on the Database Configuration Assistant that says "The ASM instance configured on the local node is a single instance. To manage this instance as a single-instance ASM restart DBCA and select the single instance database option. To manage this instance as RAC ASM, convert it to RAC ASM first".
How I can convert it??
Thanks

See the link below for converting a single instance ASM to RAC ASM. The link is to the Oracle® Database Oracle Clusterware and Oracle Real Application Clusters Installation Guide for Linux. If your platform is not Linux, check the Install Guide for your OS.
http://download-east.oracle.com/docs/cd/B19306_01/install.102/b14203/cvrt2rac.htm#BABJDAJG

How to install SSAS Cluster instance with Always On availability group(Standalone database instance) in same windows cluster servers

We have requirement like to Install the SSAS as Cluster instance where SQL Server Database engine was installed with Always On availability group.Please help me to how configure it
Currently we have following configuration in Current steup.
Node1 and Node2 are in windows cluster
Node1 has SQL Server Database engine Instance1 as Standalone
Node2 has SQL Server Database engine Instance2 as Standalone
Instance1 and Instance2 configured for Always On availability group with Listener.
Now we have to steup SSAS instance with High availibilty. I know we have only option is to install cluster SSAS instance.
Can some one provide information below.
1. How to steup Cluster SSAS instance in this servers.
2. Is this will have any dependancy on exisiting Listner name.
3. Is this affects the availiblity groups, if SSAS instance failed over to another node.
Thanks in Advance
Sriram

You will need to have SSAS installed as a clustered instance with shared storage. Refer to the whitepaper from this MSDN article
How To Cluster SQL Server Analysis Services
Availability Group is in it's own Role/Resource Group. When you create the clustered SSAS, it will create its own Role/Resource Group. This means that it will require its own virtual network name and virtual IP address and will not affect the existing Availability
Group. You can also have it on the existing Availability Group if you want to. However, you need to decide if you want SSAS to failover with the Availability Group or not. Your design choices will depend on that decision.
Edwin Sarmiento SQL Server MVP | Microsoft Certified Master
Blog |
Twitter | LinkedIn
SQL Server High Availability and Disaster Recover Deep Dive Course

Single SOA Suite Install with multiple oc4j instances and java processes

We right now have 5 BPEL processes and 5 ESB processes all running under one java.exe process. We would like to seperate some of
them out into their own java.exe processes without having to install more
%ORACLE_HOME% instances of SOA Suite. I can create an oc4j instance but of
course it doesn't have any SOA Suite stuff deployed to it. I tried to see what
the install would do with this new oc4j instance but it wants to create a new
%ORACLE_HOME% with an entire installation of SOA Suite.
Is there some sort of way to clone oc4j instances that have SOA Suite deployed to them so that you
don't need multiple %ORACLE_HOME% instances?
### How is this Issue Impacting Your Business ###
We really don't want to have a lot of %ORACLE_HOME% instances to have to maintain. We are
migrating projects over from our current integration server product and we'll
have potentially dozens more BPEL and ESB projects. We definitely want to
group and isolate projects so that outages of one project do not bring down
others that are unrelated.
We are currently experiencing periodic problems with one BPEL project that requires recycling but all the other BPEL and ESB
projects get recycled also. If we could put this project into it's own java
process without creating another SOA SUITE instance, it would be a big
help.
ANSWER
=======
You can create multiple domains in BPEL or create multiple systems/groups in ESB to group different projects.
MY REPLY:
=========
We have been using systems/groups in ESB but they all run under the save java.exe process. I would assume that having a seperate domain in BPEL would also run in that same java.exe process.
Right now, the one BPEL project we have a problem with will gobble up all the JDBC connections from time to time and that requires a recycle of SOA Suite, which means all BPEL and ESB projects that run in that java.exe process get recycled also. We're working that issue in a different ticket.
It would be nice if the SOA Suite installation would install against a new oc4j instance and not assume it has to create a complete %ORACLE_HOME% instance. The components of SOA Suite seem to be J2EE based components.
Scenario: I already have an oc4j instance called oc4j_soa and a complete %ORACLE_HOME% installation of soa suite. I then create a new oc4j instance from Enterprise Manager. Then I would deploy the esb-dt, esb-rt, orabpel, etc. components of SOA Suite to that new oc4j instance and modify the necessary config file so that it can work with OHS and the SOA Suite Databases. Is this possible?
Does anyone have any experience with this or do people typically install multiple complete installation of SOA Suite with mulitple Oracle Homes?

Hi,
yes, on metalink you get in touch with real experts....
You have to install serveral application servers to get different ORACLE_HOMEs.
For each one, you can install a BPEL PM.
But: For each BPEL PM you need your own database instance, or you have to configure them as a clustered BPEL installation.... (but i do not know if this work with non RAC DBs)

SQL2008R2 new named instance in existiong cluster as a new resource.

Hello everyone.
I'm trying to find out the best way to install a new named instance of a SQL2008R2 server in clustered environment.
The current windows cluster is a 2 node cluster and contains the DTC and 2 Named SQL Server instance:
MSSQLSERVER(Default) Network name: Clust1
Example1 Network name: Clust2
I need another instance but i want to reach it through the MSSQLSERVER's SQL Network Name.
In the end I need to be able to connect to the default instance with Clust1 and the new instance with Clust1\New.
Is it even possible to install SQL server not as a cluster service, more like inside a service as a new Resource?

Hello,<o:p></o:p>
Thank you for your reply.<o:p></o:p>
I want to create the installation on both nodes,
so when a failover occurs on the Clust1 SQL cluster, both the default and
the new instance fails over and starts up at the new node.<o:p></o:p>
I tested the standalone installation method, the
problem is that it won’t show up as a resource in the cluster service.
I have to add it manually?
No,what you are suggesting cannot be attained by standalone installation you need to add new instance in cluster aware mode.That would be your third instance in cluster which would require new disk ,IP for data files and Virtual name respectively.MSDTC can
be shared.
Please mark this reply as the answer or vote as helpful, as appropriate, to make it useful for other readers

Configure SSL SQL Server 2012 Cluster Instance

Hi,
I am trying to configure a SSL Certificate on a named clustered instance but i can't
This is the errorlog message
2014-12-11 14:28:51.49 spid10s Error: 17182, Severity: 16, State: 1.
2014-12-11 14:28:51.49 spid10s TDSSNIClient initialization failed with error 0xd, status code 0x38. Reason: An error occurred while obtaining or using the certificate for SSL. Check settings in Configuration
Manager. The data is invalid.
2014-12-11 14:28:51.49 spid10s Error: 17182, Severity: 16, State: 1.
2014-12-11 14:28:51.49 spid10s TDSSNIClient initialization failed with error 0xd, status code 0x1. Reason: Initialization failed with an infrastructure error. Check for previous errors. The data is invalid.
2014-12-11 14:28:51.49 spid10s Error: 17826, Severity: 18, State: 3.
2014-12-11 14:28:51.49 spid10s Could not start the network library because of an internal error in the network library. To determine the cause, review the errors immediately preceding this one in the error
log.
2014-12-11 14:28:51.50 spid10s Error: 17120, Severity: 16, State: 1.
2014-12-11 14:28:51.50 spid10s SQL Server could not spawn FRunCommunicationsManager thread. Check the SQL Server error log and the Windows event logs for information about possible related problems.

Hi Jame,
According to your description, you encountered the error when you were trying to configure a named clustered instance to use SSL certificate.
Based on my research, the error 17182 "TDSSNIClient initialization failed with error 0xd, status code 0x38" occurs specifically because of the fact that the string under Certificate value cannot be properly converted back to a valid thumbprint
of the certificate.
A common root cause for these symptoms is an invisible character that may have been inadvertently added to the certificate's Thumbprint value, when it gets copied out of the Certificates snap-in's rich-edit control in MMC. For example, if the thumbprint
contains the Unicode character, then it may cause this issue, the following similar issue is for your reference:
http://blogs.msdn.com/b/sqljourney/archive/2013/07/23/when-using-ssl-sql-failover-cluster-instance-fails-to-start-with-error-17182.aspx
For more details and the resolution about this issue, please refer to the following article:
https://support.microsoft.com/kb/2023869?wa=wsignin1.0
If you have any question, please feel free to let me know.
Regards,
Jerry Li

Dedicated network for AlwaysON replication traffic when a replica is a Failover Cluster Instance

Hi,
We are planning to setup dedicated network for our Availability Group replication traffic. We have a Failover Cluster Instance as the primary replica and a standalone SQL server instance as the secondary.
I understand that we will need to manually configure the database mirroring endpoints on both the replicas to listen on the specific IP.
But how do I configure the database mirroring endpoint on the Failover Cluster Instance ?
Please help.
Thanks and Regards,
Jisha

If you have a dedicated network for your Availability Group replication traffic between the FCI and the standalone instance, you need to identify if there will be other network services included in the mix. For example, your public network is already using
it's own DNS server by virtue of Active Directory integration. Your dedicated network for replication traffic may or may not have its own DNS server so configuring the endpoints would involve using either IP addresses like the one highlighted in the
blog post or using hosts file with fully qualified domain names so you can use them when creating the endpoints
Edwin Sarmiento SQL Server MVP | Microsoft Certified Master
Blog |
Twitter | LinkedIn
SQL Server High Availability and Disaster Recover Deep Dive Course

How can I convert an array with varying length into a cluster

Hi,
I need to convert an array of n elements into a cluster. Usually, it only involves the "Array to Cluster" function. But since the array length is not constant and the number of elements in the function is constant and can not be changed programaticaly, I need to find another way to do that.
Any ideas?
Thanks.

Hi,
I also tried to do this but without success. I think that's not possible.
For me it's a general problem using labview. I often wanted to generate standard controls dynamically which is not possible (or I haven't found the solution yet). So I think that's the reason why you can't create a cluster during runtime (the program would have to add controls dynamically to the cluster).
Do you have an upper limit for your array-size? Although it needs a lot more memory, it would be a possibility to create the cluster with the maximum number of elements.
If you don't have a maximum, you will have to look for another way, I'm afraid.
Thomas
Using LV8.0
Don't be afraid to rate a good answer...

Add an instance to a cluster

          Hi,
          I tried to add an new wls instance to a cluster, but after I added the new one,
          all the triffic goes to the new one. I have to restart the old instances. After
          I restart everything is fine. I am in wls7.0sp1 with IIS proxy. Thanks


          The IIS plug-in is supposed to do round-robin. I assume your settings have other
          servers and if so it sounds like a bug. LEt's know what you hear.
          S
          "Jen" <[email protected]> wrote:
          >
          >Hi,
          >
          >I tried to add an new wls instance to a cluster, but after I added the
          >new one,
          >all the triffic goes to the new one. I have to restart the old instances.
          >After
          >I restart everything is fine. I am in wls7.0sp1 with IIS proxy. Thanks

How to add a java dialog instance to Java central instance of SAP CRM

Hi Gys,
I have installed a new dialog instance for the CRM Java landscape for the version 7.01 .
CRM Abap instance is seperate and its data accessed by Jco connection by central instance but not dialog instance . Could you please let me know if any configuration steps to be followed.
Thanks in advance
Sudhakar

Could you describe your problem a little bit more.
I have installed a new dialog instance for the CRM Java landscape for the version 7.01 .
CRM Abap instance is seperate
Ok, sounds like 2 Java instances and one ABAP instance.
and its data accessed by Jco connection by central instance but not dialog instance.
This is where I get confused.
Do you have one or more ABAP instances?
What is your actual problem (what do you want to do)?

Web modules don't get deployed to cluster instance directory

Hi,
I have a domain with one cluster, which has one instance. Sometimes (!), the web aplications I deploy don't get unpacked in the directory {instanceRoot}/applications/j2ee-modules but only in {domainRoot}/applications/j2ee-modules. When that happens, they don't work and I cannot access them through their URL.
I deploy them using the web admin console, enable the to the cluster in question. Everything goes ok (or at least the UI says so) and in the domain.xml I have this:
<web-module availability-enabled="false" context-root="/myapp" directory-deployed="false" enabled="true" location="${com.sun.aas.instanceRoot}/applications/j2ee-modules/myapp" name="myapp" object-type="user"/>
<server...
<application-ref disable-timeout-in-minutes="30" enabled="true" lb-enabled="false" ref="myapp"/>
<cluster...
<application-ref disable-timeout-in-minutes="30" enabled="true" lb-enabled="false" ref="myapp"/>
In both the apps the were copied to the instance applications dir and in the ones that remained only in the domain applications dir (and do not work). In the latter, the location dir does not exist.
Anyone has gone through any similar situation? This deployment process is a bit confusing for me, wrt instances, clusters, domains etc.
Thanks!
Jo�o.

Thanks for the answer.
But how is this mechanism controlled? Sometimes, the domain just doesn't seem to propagate the changes to the cluster instances (or stand-alone instances). For example, sometimes when I change a property of a data source through the admin gui of the domain, it does not get propagated to the desired instance config, even after restarting the whole system.
I think that it may have something to do with the manual changes that I have to do sometimes in the domain.xml file (this is neede when you have huge config changes to be replicated in many servers).
Is there any option to enforce the replication of the config from the domain to the instances? Or at least to make it verbose so that I can realize what is going on underneath.
Thanks.

Re: Throughput drops dramatically when I increase the number of instances in a cluster

Thanks for the quick reply.
          Yes, we are running WLS with the performance pack.
          "Bernie Wong" <[email protected]> wrote in message
          news:[email protected]...
          > Fellow clusterers ....
          >
          > I have an EJB driver program that pumps work into a WLS cluster. For
          > scalability, I increase the number of drivers. The drivers have a big pool
          > of work to pump into the cluster.
          >
          > I noticed that with 6 (nothing special about that number) drivers, I can
          get
          > 500K worked items per hour in a single instance (clustered). That is, the
          > instance is cluster enabled and is watching for multicast, etc..
          >
          > When I add the second instance into the cluster, the cluster-wide
          throughput
          > with the six drivers drops to 370K items per hour. When I add the third
          > instance, the cluster-wide throughput drops further to 250K items per
          hour.
          > When I kill an instance, the throughput grows back to 370K items per hour.
          > When I kill another instance (to get back to one instance), the throughput
          > grows back to 500K/hr.
          >
          > Can somebody point out what we are doing wrong - do we need to tweek some
          > WLS parameters, do we have to avoid some features that limit cluster
          > scalability, etc..
          >
          > My WLS configuration is .....
          > - WLS 5.1, SP5
          > - JDK 1.2.1_04c with JIT, no Hotspot
          >
          > My hardware/software configuration consists of:
          > - an E4500 - 8 processor database server - this is only running at 15%
          > server busy
          > - three E4500 - all 8 processors/6GB RAM - application processors
          >
          > The cluster is implemented with one instance per application server.
          > Each application server is only 10-15% busy.
          >
          > My driver and application is ....
          > - stateless EJB driver and stateless EJB bean
          >
          > Your thoughts and comments would be greatly appreciated.
          >
          > Bernie
          >
          >


Thanks. It works. We screwed up the settings.
          "Bernie Wong" <[email protected]> wrote in message
          news:[email protected]...
          > Thanks for the reply.
          >
          > We ran the benchmark again and got the same scaling profile. This
          afternoon,
          > we are going to run a benchmark with a really simple driver + bean (a do
          > nothing bean) and see if we get the same scaling profile. From all the
          > responses, it would appear that WLS can scale therefore we want to isolate
          > the problem to either our code or our configuration (or both).
          >
          > I will let you know.
          >
          > Bernie
          >
          > "Rob Woollen" <[email protected]> wrote in message
          > news:[email protected]...
          > > Make sure that this was set on the client not the server.
          > >
          > > -- Rob
          > >
          > > Bernie Wong wrote:
          > > >
          > > > Jim,
          > > >
          > > > Thanks for the suggestions. One of my performance engineers ran
          several
          > > > experiments last week where he set the executeThreadCount to 6 and to
          10
          > > > along with a percentSocketReaders to 50. He saw no improvement. We
          will
          > > > rerun that experiment again since that seems to be the most common
          > > > recommendation.
          > > >
          > > > Bernie
          > > >
          > > > "Jim Zhou" <[email protected]> wrote in message
          > > > news:[email protected]...
          > > > > One possible reason is not enough socket reader execute threads, try
          > to do
          > > > > this:
          > > > > "-Dweblogic.system.executeThreadCount=10"
          > > > > "-Dweblogic.system.percentSocketReaders=50" on your client JVM
          startup
          > > > > command line.
          > > > >
          > > > > http://www.weblogic.com/docs51/admindocs/tuning.html#tuning clients
          > > > >
          > > > > WebLogic RMI
          > > > > If you are using WebLogic RMI clients and there are more than 2
          > Weblogic
          > > > > servers in a cluster, you may encounter a significant performance
          > > > > degradation (very long round trip times for stateless session beans,
          > for
          > > > > instance). The solution is to make some property changes on the
          client
          > > > side
          > > > > as explained below.
          > > > >
          > > > > The solution to this problem is to ensure that there are at least as
          > many
          > > > > socket reader threads as there are connections to the server and
          also
          > > > > allowing for some extra threads for processing other tasks. This is
          > > > > accomplished by starting the client with the command line argument
          > > > > "-Dweblogic.system.percentSocketReaders" set to a sufficiently high
          > > > > percentage (say 50) and by ensuring that there sufficient number of
          > > > execute
          > > > > threads for other processing on the client. A metric of twice the
          > number
          > > > of
          > > > > execute threads as there are servers in a cluster should work fine
          if
          > the
          > > > > above percentage is at 50. The command line argument affecting the
          > number
          > > > of
          > > > > execute threads is
          > > > >
          > > > > "-Dweblogic.system.executeThreadCount".
          > > > >
          > > > > For instance, we could use:
          > > > >
          > > > > "-Dweblogic.system.executeThreadCount=10"
          > > > >
          > > > > and
          > > > >
          > > > > "-Dweblogic.system.percentSocketReaders=50"
          > > > >
          > > > > when testing with 3 or 4 servers in a cluster.
          > > > >
          > > > >
          > > > > Bernie Wong <[email protected]> wrote in message
          > > > > news:[email protected]...
          > > > > > Thanks for the quick reply.
          > > > > >
          > > > > > Yes, we are running WLS with the performance pack.
          > > > > >
          > > > > >
          > > > > > "Bernie Wong" <[email protected]> wrote in message
          > > > > > news:[email protected]...
          > > > > > > Fellow clusterers ....
          > > > > > >
          > > > > > > I have an EJB driver program that pumps work into a WLS cluster.
          > For
          > > > > > > scalability, I increase the number of drivers. The drivers have
          a
          > big
          > > > > pool
          > > > > > > of work to pump into the cluster.
          > > > > > >
          > > > > > > I noticed that with 6 (nothing special about that number)
          drivers,
          > I
          > > > can
          > > > > > get
          > > > > > > 500K worked items per hour in a single instance (clustered).
          That
          > is,
          > > > > the
          > > > > > > instance is cluster enabled and is watching for multicast, etc..
          > > > > > >
          > > > > > > When I add the second instance into the cluster, the
          cluster-wide
          > > > > > throughput
          > > > > > > with the six drivers drops to 370K items per hour. When I add
          the
          > > > third
          > > > > > > instance, the cluster-wide throughput drops further to 250K
          items
          > per
          > > > > > hour.
          > > > > > > When I kill an instance, the throughput grows back to 370K items
          > per
          > > > > hour.
          > > > > > > When I kill another instance (to get back to one instance), the
          > > > > throughput
          > > > > > > grows back to 500K/hr.
          > > > > > >
          > > > > > > Can somebody point out what we are doing wrong - do we need to
          > tweek
          > > > > some
          > > > > > > WLS parameters, do we have to avoid some features that limit
          > cluster
          > > > > > > scalability, etc..
          > > > > > >
          > > > > > > My WLS configuration is .....
          > > > > > > - WLS 5.1, SP5
          > > > > > > - JDK 1.2.1_04c with JIT, no Hotspot
          > > > > > >
          > > > > > > My hardware/software configuration consists of:
          > > > > > > - an E4500 - 8 processor database server - this is only
          > running at
          > > > > 15%
          > > > > > > server busy
          > > > > > > - three E4500 - all 8 processors/6GB RAM - application
          > processors
          > > > > > >
          > > > > > > The cluster is implemented with one instance per application
          > > > server.
          > > > > > > Each application server is only 10-15% busy.
          > > > > > >
          > > > > > > My driver and application is ....
          > > > > > > - stateless EJB driver and stateless EJB bean
          > > > > > >
          > > > > > > Your thoughts and comments would be greatly appreciated.
          > > > > > >
          > > > > > > Bernie
          > > > > > >
          > > > > > >
          > > > > >
          > > > > >
          > > > >
          > > > >
          >
          >

Throughput drops dramatically when I increase the number of instances in a cluster

Fellow clusterers ....
I have an EJB driver program that pumps work into a WLS cluster. For
scalability, I increase the number of drivers. The drivers have a big pool
of work to pump into the cluster.
I noticed that with 6 (nothing special about that number) drivers, I can get
500K worked items per hour in a single instance (clustered). That is, the
instance is cluster enabled and is watching for multicast, etc..
When I add the second instance into the cluster, the cluster-wide throughput
with the six drivers drops to 370K items per hour. When I add the third
instance, the cluster-wide throughput drops further to 250K items per hour.
When I kill an instance, the throughput grows back to 370K items per hour.
When I kill another instance (to get back to one instance), the throughput
grows back to 500K/hr.
Can somebody point out what we are doing wrong - do we need to tweek some
WLS parameters, do we have to avoid some features that limit cluster
scalability, etc..
My WLS configuration is .....
- WLS 5.1, SP5
- JDK 1.2.1_04c with JIT, no Hotspot
- we enabled performance pack
My hardware/software configuration consists of:
- an E4500 - 8 processor database server - this is only running at 15%
server busy
- three E4500 - all 8 processors/6GB RAM - application processors
The cluster is implemented with one instance per application server.
Each application server is only 10-15% busy.
My driver and application is ....
- stateless EJB driver and stateless EJB bean
Your thoughts and comments would be greatly appreciated.
Bernie

The short answer is to set the following tuneables on the client side:
java -Dweblogic.system.executeThreadCount=10
-Dweblogic.system.percentSocketReaders=50 <your client program>
http://www.weblogic.com/docs51/admindocs/tuning.html#tuning clients
Srikant, [email protected], http://weblogic.bea.com/, etc.
Bernie Wong wrote:
Fellow clusterers ....
I have an EJB driver program that pumps work into a WLS cluster. For
scalability, I increase the number of drivers. The drivers have a big pool
of work to pump into the cluster.
I noticed that with 6 (nothing special about that number) drivers, I can get
500K worked items per hour in a single instance (clustered). That is, the
instance is cluster enabled and is watching for multicast, etc..
When I add the second instance into the cluster, the cluster-wide throughput
with the six drivers drops to 370K items per hour. When I add the third
instance, the cluster-wide throughput drops further to 250K items per hour.
When I kill an instance, the throughput grows back to 370K items per hour.
When I kill another instance (to get back to one instance), the throughput
grows back to 500K/hr.
Can somebody point out what we are doing wrong - do we need to tweek some
WLS parameters, do we have to avoid some features that limit cluster
scalability, etc..
My WLS configuration is .....
- WLS 5.1, SP5
- JDK 1.2.1_04c with JIT, no Hotspot
- we enabled performance pack
My hardware/software configuration consists of:
- an E4500 - 8 processor database server - this is only running at 15%
server busy
- three E4500 - all 8 processors/6GB RAM - application processors
The cluster is implemented with one instance per application server.
Each application server is only 10-15% busy.
My driver and application is ....
- stateless EJB driver and stateless EJB bean
Your thoughts and comments would be greatly appreciated.
Bernie

Converting a standalone SOA/AIA instance to a cluster instance.

Similar Messages

Maybe you are looking for