The basic layout consists of two or more physical ESXi hosts. We used redundant blade server enclosures with adequate processor and memory specs and fiber channel storage devices. Every server that was needed for the DCS was installed as a Windows 2008 virtual server. Servers were broken up into pairs and installed on resources in such a way that the system was load balanced and fault tolerant. This setup would allow us to survive any single hardware failure without interrupting the normal operation of the plant area.

We established two VMware View servers, one to work as a replica of the other. Most of the operator stations around the mill have at least two machines to work from. Often these machines are running a quad monitor setup giving the operators eight screens to display graphics and trends. We decided to separate each pair of machines into two groups. Each group connected to a separate physical network, a separate VLAN, and each routed to a unique View server (each View server also ran DHCP as well as a thin client management application). We used the tag functionality of VMware View to segregate which pools a given View server would allow access to. This allowed us to further separate the virtual machines into groups. This essentially gave us two separate systems running side‐by‐side, but allowed us to manage them through one View Server Connection Group. 

- Renee Robbins Bassett, Managing Editor

Link to the full version of this white paper at  bit.ly/awtech001

This white paper was created by  Global Process Automation. Publication  Date: January 2012