Do check the comments for clarification.
I have a BIG DOUBT (appears big to me) and my only intention of writing this post is to get a clarification. So, if you are looking for some knowledge gain over here, you may instead be inviting a BIG PAIN in your head. I am inviting only contributors who can help me clarify MY BIG DOUBT. Anyone else reading this article will be doing so at his own risk.
Okay. Let me start now.
I was reading the first chapter, ECM Basics of Pawan Kumar’s book on Documentum Content Management Foundations and came across the following.
The term Content Server is used in two contexts – the Content Server software that is installed and resides on the file system and the Content Server instance, which is the running process that resides in memory and serves content at runtime. However there is little chance of confusion since the usage is often clear from the context and the term Content Server is typically used without additional qualification (software or process/instance).
A Content Server instance is dedicated to and manages only one repository. However, we will see later in architecture discussion that multiple Content Server instances can be dedicated to the same repository. This is typically done for performance reasons where the multiple Content Server processes divide up the load for serving content from the same repository.
Pawan is pretty clear in what he has mentioned. I got the doubt while relating it to the practical usage.
In practice while creating an environment setup we first install Content Server on a server (after installing database) system and then we create repositories. We can have multiple repositories. What we have now is one installation of Content Server Software and more than one repository.
Once all the installation and configuration process is over (including Content Server, Application Server, Webtop/Custom Application) and the application is ready for use, we can connect to the application through a browser. We get a login screen where we get the drop-down having the repository names to choose from. We can select any repository on the login screen and connect to it after authentication.
Now the thing which I am wondering about: As per the book the repositories have dedicated Content Server Instances. So if we have two repositories, we should as well be having two Content Server Instances which are actually running Processes/Programs. Where can I see these instances? When are they created? How can I make two Content Server Instances to serve single repository?
I don’t have an answer to these questions.
After a lot of brain storming I could think of starting/stopping of docbases using Documentum Server Manager. By starting a docbase we are indeed starting a process or a program. This program for sure is dedicated to the particular repository. So, shall it be correct if I say that by starting a docbase through Server Manager, I am actually starting a Content Server instance?
….BOOOOM….
I am sorry if it is a NON-SENSE of a very HIGH LEVEL, but this was the best I could think of. Even if I assume the statement to be true, how am I going to make an additional Content Server instance to serve the same repository? This is creating a much bigger pain in my head. I am confused and I am sure I am creating pain in my audience’s head too. Just wondering whether docbroker is also involved somewhere in the scene?
Let me stop here. This pain is becoming unbearable and I want to get rid of it.
Looking forward for some clarification. Any kind of help in this regard will be highly appreciated.
Have you an account with EMC Powerlink site?
If so please refer to this thread for one of the best of explanations by Fabian Lee (EMC Employee)
https://forums.emc.com/forums/thread.jspa?threadID=47050
If not, here id what posted there…..(I am not posting the entire stuff…)
But this should be a good remedy for your headache !!
……..
………
It sounds like you are trying to implement the most common configuration for High-Availability (HA) deployment. A single repository, backed with a pair of Content Server /Docbroker installations. If one installation goes down, clients will be automatically redirected to the other.
I think the most difficult concept in this setup is making the delineation between a CS installation and the repository. Because when we are in an non-HA environment, the term “docbase”, “repository”, and “content server” are all interchangeable. “Repository” is the collection of content+metadata that we typically think of as the docbase. “Content Server” denotes a single instantiation of the repository, and a repository can have multiple CS installations. Each shares the same database tablespace.
So when we setup up this particular HA configuration, the second CS installation should not create its own database tablespace. Instead, use the same tablespace and credentials you used when creating the first CS.
Once you have both CS installations up, go into the server.ini on each server and add both docbrokers . Use the projection target sections, as shown below.
[DOCBROKER_PROJECTION_TARGET]
host =
port = 1489
proximity = 1
[DOCBROKER_PROJECTION_TARGET_1]
host =
Finally, it doesn’t make any sense to make your DCTM HA if you don’t provide this service to your clients. Set the primary and backup docbrokers on your application server or client, see this thread where I describe this to another poster.
https://forums.emc.com/forums/thread.jspa?threadID=47018&tstart=50
As I describe in the above post, you should put a load balancer in front of a pair of web servers if you want to provide HA to your end users.
HOPE IT HELPED….
So basically this is what you do make 2 CS serve a single repository:
On the second machine after you install the content server binaries, run the repository configuration program. On the config program choose to create a new docbase but give the same details (i.e. docbase name, docbase id, database details)-[which is mentioned above as sharing the same tablespace] as the first server . This will enable you to use 2 content servers for the same docbase.
Lakshminarayanan Rengarajan
Uttkarsh,
Your thoughts have been mostly logical. Yes you are starting a content server instance when you “start a repository”. On Windows, you can see them listed in the Services as well. If multiple content server instances are started, you can see them under processes on Task Manager as well. On Linux or Unix systems you could list processes.
Setting up multiple content server instances for one repository is an advanced concept. The installation process is partly similar to creating a repository. Typically the multiple CS instances for one repository reside on separate machines.
If you have questions related to the book or preparation for E20-120 you can use the discussion forum at http://dmcram.org
Hope this helps,
Pawan
Tons of thanks to both Pawan and Rengarajan.
Your clarifications are indeed a great remedy for my headache.
It is great to get a clarification from the Author himself.
Now I am able to relate the theory with the practice.
Thanks a lot!!!
[...] 28, 2009 by uttkarsh Ultimately after four and a half months I got an opportunity to implement the Single Repository Multiple Content Server model. We had to [...]
Hi Utkarsh,
I can’t able to access your article on “Single repository Multiple Content Servers”.
Is there any other link to get it.
Please advice me. I am really looking for this info.
Thanks,
Prasad
Hi Prasad,
I am sorry for the delayed reply. I had kept that post private because somehow I felt that it was not complete. I am turning it public with few changes. Hope it’s of some help to you.
Uttkarsh
Hi,
I am very new to documentum .
I registered with this EMC2 for downloading documentum… and it alllows the access till documentation … but its not allowing me to the download center.
Can any one help me plz.
Thanks,
Manju
In order to download software from EMC Subscribenet you need to subscribe for it.
Alternatively you can better download Documentum Developer Edition which is available for free at https://developer-content.emc.com/downloads/documentum_dev_edition.htm