Are people using Websphere Enterprise or Websphere ND version with CLM?
Hi,
We're currently using Tomcat as our App Server for all our repositories. We're starting to test out the usage of Websphere as a replacement for Tomcat. Currently we've been able to install CLM on WAS with SSO enabled, which is a key feature driving this. The initial install was done with the standard WAS Enterprise edition, but now we're looking into WAS ND as an option. The setup has been a lot more complicated and I'm curious if people are using the standard Enterprise Edition or are they using WAS ND in their environment. And why did you choose that one? Thanks, |
4 answers
First a disclaimer. I am part of CLM support team.
I prefer ND for distributed environments for the following reasons: 1. There is one AdminConsole that is used to manage all the appservers 2. If a proxy is being used, then there is no need to manually merge the plugin-cfg.xml files. 3. A server template can be created with the basic configuration for jvm. thread pools and all the needed custom properties. So when a new server needs to be created, the server template can be used. So all that would be needed is to customize the value for JAZZ_HOME and log4j properties. In short, ND provides ease of management. Comments
Michael Walker
commented Sep 04 '13, 7:18 p.m.
Thanks for the reply. I'm following the instructions documented here, https://jazz.net/library/article/1123, except we're only using 2 servers. Server1 has JTS installed as well as DB2 for the dbs. Server2 has CCM and RQM installed. WAS ND is installed on both.
|
I think one more really important point for WAS ND is reliability and failure safety.
If you want to enable those features, e.g. if one WAS is killed that another one takes his part immediately you need WAS ND. Same for load balancing. Greetings, Simon Comments
Michael Walker
commented Sep 04 '13, 7:20 p.m.
Thanks for the reply. The feature to recover from a WAS getting shutdown is another feature we're looking into.
|
It sounds like when ND was installed, the option to also create a profile during the install was selected. Here are the steps I would take:
Here are the assumptions I am using when setting up a production environment 1. Need 24/7 up time 2. Need to ensure that if one server has a catostrophic failiar it does not affect the other servers Topology have each process running on its own host. This means Dmgr will be running on one server, jts on another, ccm on another and so on. Steps to take. 1. Install Nd on each node but unselect the option to create a profile 1a. I do this because if the default settings are used then ND will use the non descriptive names for the servers such as Dmgr01, app01, ap02, server1, server2 and so on 2. On the server where the Dmgr will be running, start up the profile management tool, select create profile and then select advanced. using the advanced option, you can modify the profile and node names Creating the application servers 1. start the profile management tool 2. for the type of profile there are two options, apServer and custom appserver. 2a. If appServer is created, the default server instance with the name server1 will be created 2b. If custom appServer is selected, then no server instance will be created initially 3. I would select the custom application server. The reason for this will become apparent later 3a. When creating the custom profile, make sure it is federated to the Dmgr. This is so the Dmgr is aware of this profile Creating the server instances 1. First create a server instance which will be used as a template. 1a. Modify the jvm settings, and all other custom properties that will be used by all the clm applications 2. create clm server instances . During the wizard, you will have the option to select the template created in step 1 2a. Also during the server creation, you can specify unique names, such as jts_server, qm-server and so on. At this point you will have CLm infrastructure in place and all that is needed is to deploy and configure CLM. |
oh yeah, and lets not forget the DB which should be on its own host as well.
|
Your answer
Dashboards and work items are no longer publicly available, so some links may be invalid. We now provide similar information through other means. Learn more here.