Test
Infrastructure: Web Content Management End to End:
Authoring, Rendering, Library import/export
and Syndication
Abstract
The purpose of this document is to outline the steps by which the Web Content
Management System Verification Test (SVT) team installed, configured, and
tested Web Content Management. Authoring, Rendering , API and Syndication
were tested in this scenario.
Content Introduction
The environment included a rendering cluster and an authoring cluster with
a staging node in between. Content was moved between the different environments
with automatic syndication. The software versions used in this test environment
are as follows:
· WebSphere
Application Server 7.0.0.0
· WebSphere
Portal 6.1.0.1
· Lotus Web
Content Management 6.1.0.1
· DB2 9.5.0.1
· Oracle 11.1.0.6.0
· IBM HTTP
Server 7.0.0.0
· IBM Directory
Server 6.1
Because 64-bit support is new in this release of WebSphere Portal, all
of the operating systems, databases and application server versions used
in this test scenario were 64-bit.
The following test environment was configured to test the scenario.
test_wcm.jpg
Server Specifications
The following table lists the specifications for the computers used in
the infrastructure.
|
Machine
|
OS
|
# of Cores
|
CPU Speed
|
CPU Type
|
RAM (GB)
|
Function
|
|
IBM Blade HS21
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
Authoring Cluster
|
|
IBM Blade HS21
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
Authoring Cluster
|
|
IBM Blade HS21
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
DB2 Server
|
|
IBM Blade HS21*
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
1.5
|
Dmgr / Httpd
|
|
IBM Blade HS21
|
Solaris 10
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
Rendering Cluster
|
|
IBM Blade HS21
|
Solaris 10
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
Rendering Cluster
|
|
IBM Blade HS21
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
Oracle Server
|
|
IBM Blade HS21*
|
RHEL 5.1
|
8
|
2.33Gz
|
Intel Xeon E5345
|
1.5
|
Dmgr / Httpd
|
|
IBM Blade JS21
| AIX 5.3 TL 7
|
4
|
2.5Ghz
|
IBM PowerPC
|
4
|
Staging Cluster
|
|
IBM Blade JS21
|
AIX 5.3 TL 7
|
4
|
2.5Ghz
|
IBM PowerPC
|
4
|
Staging Cluster
|
|
IBM Blade HS21
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
DB2 Server
|
|
IBM Blade HS21*
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
1.5
|
Dmgr / Httpd
|
|
IBM Blade HS21
|
RHEL 5.1
|
8
|
2.33Ghz
|
Intel Xeon E5345
|
8
|
LDAP Server |
* The Deployment Manager/ HTTP Server machines
were hosted as VMware images, on the same host hardware as the other Intel
machines.
Installation and Configuration
WebSphere
Portal 6.1 Information Center
Please refer to the topics listed in the steps below for more detailed
instructions on the steps to install and configure the environment used
for this test.
The environment was installed with the following steps:
1.0 Install DB2 Server, using the topic “Planning for DB2 on UNIX or Linux”
a. Create databases and domains.
2.0 Install WebSphere Portal Network Deployment Manager, using the topic
“Preparing the WebSphere Application Server Network Deployment on AIX”
3.0 Install the Web server, using the topic “Preparing a remote Web server
on AIX/Linux/Solaris”
4.0 Install and Setup the primary node, using the topic “Preparing the
primary node on AIX/Linux/Solaris”
a. Install WebSphere Portal
b. Configure/Transfer the remote
database(s)
c. Setup Cluster
d. Configure primary node to communicate
with the network deployment manager
e. Configure primary node to display
WCM data through Web Server
f. Configure cluster to use external
Web Server
5.0 Add an additional node to the cluster, using the topic “Preparing
additional nodes on AIX/Linux/Solaris”
a. Install WebSphere Portal
b. Add the node to the cluster.
6.0 Configure Content Server, using the topic “Setting up a content server”
7.0 Create WCM Libraries/Content for Virtual Portals, using topic “Creating
Web content”
a. create the content
b. specify the user level permissions for the
content
8.0 Create PZN Rules, using topic “Personalizing your content”
9.0 Tune All Servers for Portal and WCM related settings.
10.0 Restart environment and verify all settings
Test / User Configuration
The following tests were conducted using an internal IBM too that is similar
to LoadRunner, to simulate the multiple users performing their various
tasks over specific periods of time.
Authoring
The authoring use case created content in a single library, which has three
authoring templates, one site, three siteareas and one 5 stage workflow
(Draft, Approval, Published, Expired and Reject).
Two types of tests were completed, a reliability test and a stress test.
The reliability test was for 70 Virtual Users with 20-60 second think time.
The stress test was for 30 Virtual Users with 0 – 20 second think time.
Three distinct runs were attempted, one with a basic workflow (draft ->
approve ) , one with a more complex workflow ( draft -> approve ->
published -> expired ) and one with a more complex workflow that has
custom actions. All runs were executed for 12 hours for both reliability
test and stress test. The complex run was executed for 72 hours for reliability
test.
Rendering
The rendering use case involved the creation of a complex 'Employee Portal'
type website. The site contained PZN rules, Menus and JSPs exercising the
WCM API. The contents are rendered on 20 pages, which have copies of WCM
content viewer portlet of their own.
Rendering was tested with 200 Virtual Users with a 20 – 60 second think
time over 72 hours.
Library import/export
The library import/export use case is the test against real customer data,
which has 12K items. A validation test is done to make sure the library
is imported correctly by creating contents referring to existing items.
It took about 14 hours to import or export this customer data.
Syndication
Syndication was performed after the authoring library was populated. A
syndicator/subscriber relationship was established between the authoring
cluster and staging server, and a syndicator/subscriber relationship was
established between the staging server and rendering cluster (as shown
in the diagram above). The automatic syndication interval was set to 300
seconds.
After this initial syndication setup, both the authoring and rendering
use cases were executed on their respective clusters, and it was verified
that data authored on the authoring cluster was correctly syndicated to
the staging and then to the rendering cluster. This scenario was run for
72 hours.
NOTE
: Every customer is different. Our results were obtained in a controlled
test environment. Customer environments are usually less optimal and may
not get the same results. Understanding your environment (usage scenario,
network, etc...) is crucial before recommending scaling numbers, hardware
and solutions