Application Development Software

Hi All,

We have both contribution and consumption environments, each with:

* 2 x Windows 2003 Servers, clustered, with Oracle UCM 10gR3, running a Site Studio 10gR3-based website
* Oracle Database 11gR1, SPARC hardware, Sun Solaris Zones
* HP Polyserve on SAN for shared filesystem between cluster nodes

We are looking at moving each of these physical installations to a single node environment:

* Windows 2003 Server VMs for the UCM 10gR3 installation and Site Studio 10gR3 website
* Local disk for the filesystem
* Red Hat Linux VMs for the Oracle 11g Database

We understand the support and redundancy risks with this move.

The logistics of it are proving very challenging and I would appreciate any advice on strategy.

We see the following options, with pros [+] and cons [-] of each.

1. Doing an OS 'Physical to Virtual' (P2V) snapshot of each existing Windows UCM Server into a VM
+ Snapshot state preserves exact UCM installation configuration.
+ Content in Workflow can be preserved in it's correct step.
+ Existing IDC instance name can be preserved, not breaking any external absolute links that have been made which reference the IDC name in the path.
- Content contributed after the P2V process will not be captured, therefore contribution will need to be unavailable for the duration of the process. Resumption of contribution will need to be made to the new VM.
- Existing Sun Solaris databases cannot be P2V'ed into a Linux database. Will need to build new databases and attempt to replicate the existing UCM schemas and hope they "line up" with the UCM application installation, all at the same time as the Windows P2V process = very long downtime.

2. Completely fresh UCM build.
+ Content and Configuration can be migrated with Archiver and Config Manager to ensure a seamless new UCM build.
- Content in Workflow cannot be moved with Archiver to a new instance. We have A LOT of content in Workflow at any one time.
- A new IDC name must be used. Archiver will not allow content to be moved between instances with conflicting IDC names. The WebURLMap can be used to capture any absolute URL references, e.g.
http://mydomain.com/idcname/some/path/document.pdf, and map it to, http://mydomain.com/new_idcname/some/path/document.pdf, but this won't work for any dynamic paths such as http://mydomain.com/idcname/idcplg?IdcService=DOC_INFO&dID=1234&dDocName=MyDocName

There are other issues to consider, but these are the main ones.

I would appreciate any advice or assistance from others who have attempted this process.