JCAPS Deployment Design
Hi All, I have a number of questions about the differences in the Deployment Design/Arch between JCAPS and ICAN.
1) In the ICAN world, I created multiple logical hosts in my environment, each with several Integration Servers and Message Servers. Components on one deployment profile where able to subscribe to messages on a Message Server on a different DP. It does not appear that this can still happen. Am I correct with this?
2) I understand that logicalhost are now just sort of a place holder, and the real work is done on a Domain. I've been able to successfully create domains and run simple projects through them. It seems to me that a message spends its whole life in a domain, and cannot cross between them. Meaning that a message will enter and exit JCAPS all within the same domain.
3) The correct way to build/deploy a project is to code it in eDesigner, build the .ear using the Deployment Profile, and than Deploy it using the eManager?
4) Is it true that a Domain can only have 1 Integration Server and 1 Message Server? If this is the case, why is it possible to build out multiple ISs and MSs on a logicalhost in the environment tab? I have not been able to successfully deploy a project that deploys jcds to multiple Integration Servers. I'm a doing something wrong, or is this a fact of life in JCAPS, and the answer is lots and lots of simple domains and a single LH?
5) eManager referes to 'Servers' which contain a single IS and Message Server. I assume each of these 'Servers' is a domain?
6) What does everyone think of these apparent changes?
Any help would be very much appreciated!
[1664 byte] By [
DHS15243a] at [2007-11-27 6:59:09]
# 1
In JCAPS, each logical installation allows you to create one or more domains. Each domain is a combination of a single Integration Server and a single IQ Manager. Each domain is an instance of the logical. In other words, all of the domains are at the same maintenance release level as their common logical installation.
Typically, the application EAR file is built using the Enterprise Designer. This is done by creating a deployment profile that includes one or more connectivity maps whose resources are then mapped to the external system components of a single environment as defined in the Enterprise Designer. The environment that you select when you create the deployment profile can be comprised of one or more Integration Servers and one or more IQ Managers as well as other external systems like Oracle, SAP, etc. When have finished mapping the connectivity maps to the deployment profile you then build the EAR file.
Once the EAR file is built, you can deploy it either using the Enterprise Designer or the Enterprise Manager. If you use the Enterprise Designer, it will upload the EAR file to all of the Integration Servers that were targeted in the environment specified in the deployment profile. If you use the Enterprise Manager, the EAR file will be upload to the Integration Server that you specify in the Deployer.
Applications can publish and subscribe to any IQ managers to which their corresponding JMS clients have been deployed. In other words, if you have an application that subscribes to topicInbound and publishes to topicOutbound, it is entirely possible to deploy the topicInbound->cm_jcd1 JMS client to the IQ Manager called iqmIn on port 30637 and the topicOutbound->cm_jcd1 JMS client to the iqmOut on port 30997 so long as both iqmIn and iqmOut were defined in the environment that was specified in the deployment profile.
The logicalhosts that are defined in the Enterprise Designer have absolutely nothing to do with the actual logicalhosts that are installed on your systems. They are merely containers for defining Integrations Servers and IQ Managers in the logical environment.
Each domain is an instance of a real logicalhost and is comprised of an Integration Server and an IQ Manager. If an environment in Enterprise Designer has two Integration Servers and three IQ Managers then there must be at least three domain, but could be as many as five domains. All of the domains could be created on one common logicalhost installation or on separate ones. In Enterprise Designer, they could be defined under one logicalhost or five. Regardless, the number of logicalhosts in Enterprise Designer means nothing to what actually is installated on your hardware.
Regards.
# 2
See my comments below:
2) I understand that logicalhost are now just sort of a place holder, and the real work is done on a Domain. I've been able to successfully create domains and run simple projects through them. It seems to me that a message spends its whole life in a domain, and cannot cross between them. Meaning that a message will enter and exit JCAPS all within the same domain.
>>> We do process messages between different domains in JCAPS. We currently have a main process deployed on a domain on Server-1, which publishes messages to a common error handling process deployed on a different domain on Server-2. The communication is established using a dummy JMS IQ Manager pointing to the Server-2 on which the common service is deployed.
3) The correct way to build/deploy a project is to code it in eDesigner, build the .ear using the Deployment Profile, and than Deploy it using the eManager?
>>> Another option is to use command line deployment tool 'DeployCli' instead of eManager - to deploy (using which you can deploy remotely - an option not available in ICAN)
4) Is it true that a Domain can only have 1 Integration Server and 1 Message Server? If this is the case, why is it possible to build out multiple ISs and MSs on a logicalhost in the environment tab? I have not been able to successfully deploy a project that deploys jcds to multiple Integration Servers. I'm a doing something wrong, or is this a fact of life in JCAPS, and the answer is lots and lots of simple domains and a single LH?
>>> Like I mentioned above, you might need to have multiple Message Servers in case you want to establish communication with projects deployed on other servers (with actual message servers). Similarly you need multiple ISs when a single project has components runnings on different Integration servers running on different servers for load balancing in case of high volume interfaces.
