Originally posted by: residium

Hi!

Today we have one (test/dev) server and one (prod) server. I know, that this is not optimal. We would like to implement a solution based on one server (test), one server (QA), and perhaps a clusterad server x2 (prod).Also we would like to use a shared network disk if possible.

I have found very little from IBM/mercator on how you plan and implement an architecture like this. Does anyone have some docs, links or tips to share I would be grateful.
#IBM-Websphere-Transformation-Extender
#IBMSterlingTransformationExtender
#DataExchange

Originally posted by: TimUK

We have a production environment on Windows 2000 that operates in a cluster.

In essence we install WTX launcher on each nodes local disk. All maps and systems and runtime material is then installed on the drive thats swings.

There is a setting in the .ini file to specify a cluster directory (on the swing device).
Another setting on the advance tab of the launcher admin tool specifies an RMI server address - its documented in the Infocentre.

The Event Server is then specified in the server cluster administration (MSCS in our case) and the failover criteria set. The service is alsways started and stopped from there.

It does get more complicated than that when you add other installs on the server such as MQ Series or a database as each product has its own oddities and they need to work together.

We have set up an active/passive scenario as that suits us but it has its downsides. When OS or software products need patching it may not be possible to do one node at a time. Certainly thats the case with Oracle database as the patches often affect the data dictionary which has to be stored on the shared drive.

From a performance perspective failover is pretty quick (<90secs) when at full tilt. It does depend on your requirements though. This design is not sufficent on its own if you want faster failover.

Hope that heelps
Tim
#IBMSterlingTransformationExtender
#IBM-Websphere-Transformation-Extender
#DataExchange

Originally posted by: smcasey

Regarding your production environment on Windows2000 that operates in a cluster- what are you using for a transaction manager for MQ triggers or database triggers? Is it not correct that these are not guaranteed using MSCS? Are all your triggers filed based?

There is very little from the vendor regarding active/passive or clustering other than MSCS and a very large enterprise solution (MB). Are there any other architectures that you are aware of that satisfy variety of triggers (file based, mq, database)?
#IBM-Websphere-Transformation-Extender
#DataExchange
#IBMSterlingTransformationExtender

Originally posted by: repanzer

smcasey, not sure if you noticed, but the last response is from october. Just an fyi!
#DataExchange
#IBM-Websphere-Transformation-Extender
#IBMSterlingTransformationExtender

Originally posted by: smcasey

Interested in any responses that have been implemented over the past few years on this- Windows 2000,2003 server. Interested in how you are handling transactions (transaction managment) for non-file based triggers. Thanks.
#IBM-Websphere-Transformation-Extender
#IBMSterlingTransformationExtender
#DataExchange

Originally posted by: janhess

I used to use Launcher. I believe you can use Message Broker.
#IBMSterlingTransformationExtender
#IBM-Websphere-Transformation-Extender
#DataExchange