Showing results for 
Search instead for 
Did you mean: 

EP 6.0 SP11 Installation in Cluster Environment

Former Member
0 Kudos


We are planing to install our Portal server in cluster environment for our production. It has been decided to go as per the follow procedure, but i feel we are missing some thing in this approach.

We have 4 Servers Available in our Landscape, 2 Servers are connected to SAN memory and Two servers are independent.

1. The Database and Schemas will be installed in 1 Server linked with SAN.

2. The WAS Cetnral Instance and the Portal will be installed in one of the Independent Server with host name portalprod and Virtual IP Address

3. One more WAS Central INstance and Portal will be in the Second Independent Server (Exact Replica of the First Server with same host name and Virtual IP Address)

and the 2 independent servers are linked to SAme Database.

The Microsoft Clustering is INstalled in the 2 Independent Servers for Load Balancing.

4. The Another server linked to SAN will be having the ITS.

Is this approach is correct? I guess in this there is no Java (J2EE Engine ) Clustering involved.

Can any one please suggest.

Thanks & Regards


Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos


It's not very clear if you'll be using fail-over for the DB node but I'm going to assume so. If not then your DB will be a single point of failure.

Also it is not possible to have 2 Central Instance (CI) in the J2EE cluster. You would have 1 CI ( the CI in an EP setup would be your SCS Instance (message and enqueue server) along with a Dialog Instance ( dispatcher and server node). NOTE the term CI means SCS + DI. Please note that in a fail-over scenario you do NOT have to install the DI with the SCS. All that's needed for fail-over is for the SCS to fail-over. After fail-over the Dialog Instances (DI) will re-connect to the SCS Instance.

So to scale out you could install more Dialog Instances...

So what you would have is 1 Database, 1 SCS Instance and at least 1 or more Dialog Instances...

Answers (4)

Answers (4)

0 Kudos

Hi all,

We are also trying to achieve a similar scenario. We're trying to install using MSCS to cluster only the SCS to a virtual host. This part we seem to have working. When we then try to install CI pointing to the SCS installation we get an error regarding a problem accessing the physical drive letter used in the SCS install (e:\ in our case). We are using the product_ha.catalog.

We're using Microsoft Server 2003 for all machines.

I've emailed with no response as of yet.

Anyone seen this? Any help would be greatly appreciated!


Former Member
0 Kudos


Serge is correct, you can send email to for free, for clustering/HA advice. You can also find further reading on -> SAP Web AS ->

SAP Web AS 6.40 SR1 and Related Documentation

as well as

I have sent email to, in my past two projects, and they came back with timely, and relevant information for us to deploy are architecture.

Deploying the architecture with the help of SAP Consulting services based on that note (803018) is chargeable rates.


Former Member
0 Kudos

Hi All,

Thanks very much for the info. Ihave gone through that webinar.

After that we have framed our installation like this.

1. Java CI + SCS + EP

2. Java DI + SCS

For these two servers NLB and Microsoft Cluster service will be used for Load balancing and Failover between SCS in both the servers

3. ITS & DB

4. ITS & DB in both these servers DB files will be in SAN and the NLB and Microsoft Cluster service will be used for Failover between ITS and aswell DB.

Meanwhile we have sent a mail to also for further advise.

Once again thanks for all and also if there are any further suggesions on the revised architecture please let me know.

Thanks & Regards


0 Kudos

There are specific recommendations regarding clustering/failover architecture for SAP WebAS Java 6.40 installations. The SAP recommended configurations are described in the Know How call dated January 17 which you can find under 'events' and 'SDN Webinar Schedule'. It is currently on 'project-basis' only as described in SAP note 803018.


Former Member
0 Kudos

Hi Sumanth,

He is entirely correct. However, you can have clustering in two ways:

1. have cluster over multiple machines.

2. have cluster within the same machine.

For the second scenario, you have to get a machine that has a really good configuration. More on this topic can be found at

Thanks and Regards,