When an IT department is trying to determine a server’s nomenclature, the situation can actually become quite contentious; when you roll virtualization into the fold, things can get even more complicated. I’ve settled on several design elements for naming systems, whether they are physical or virtual, servers or storage, printers or I/O devices.
The single guiding theme to a system nomenclature is to be self-documenting either at a basic level or in painful detail. The absolute basic information that I’d like from a system’s name is to be able to determine the following attributes:
- Whether the system is virtual.
- Whether the system is in development or production.
- What application or operating system is running on the system.
- Whether it is the first, second, or third (etc.) of a sequence.
These basic identifiers can create a nomenclature that works for physical and virtual systems. Figure A shows an example that does this for a few types.
Click the image to enlarge.
The obvious missing component is location. As I have been around the block a few times, I’ve determined that I am better off not having the location within the name of a server. Virtual machines move around, and I’d prefer that the burden of renaming the virtual machine not be associated with the built-in flexibility of the platform. Windows renames fine enough, but applications can require a bit of manipulation to accommodate name changes. Besides, it is just irritating when the first virtual machine shows up in a location that doesn’t have the site nomenclature.
This framework is also cluster-friendly, hence the triple digit sequencing for instances in positions 8, 9, and 10. Basically, position 8 would indicate a cluster, while leaving positions 9 and 10 for plenty of space to identify each node. As an example using the framework in Figure A, a cluster of five VMware ESXi hosts in a cluster named ESXI101 would be named as follows:
- PMPESXI101: Node one of the cluster (physical, production, ESXi)
- PMPESXI102: Node two of the cluster
- PMPESXI103: Node three of the cluster
- PMPESXI104: Node four of the cluster
- PMPESXI105: Node five of the cluster
This is just one of the ways that I have gone about this task, and I realize there are many ways to approach this topic. How do you name systems now that virtualization is a significant player in the data center? Share your comments in the discussion.