Figure 5.2 Nano Server Recovery Console
You may have heard people talk about pets vs. cattle in relation to servers, and this
conversation refers to the new ways servers are deployed and managed. Traditionally,
IT departments think of servers as pets. The servers are named, cared for, and healed
if sick. Moreover, each server has a unique purpose that would be hard to replace. In
the new world, servers are cattle. If a server is sick, it is “shot” and replaced with a new
one. Effort is not wasted trying to heal servers, and all configurations can easily be
deployed using declarative technologies like PowerShell DSC (Desired State
Configuration), Chef, or Puppet to a standard OS image. Nano Servers are the epitome
of this thinking and are designed to be created fast, deployed fast, and replaced fast as
needed.
I mentioned a smaller footprint. How much smaller? The typical size for Nano Server
is 500MB, but this will vary depending on the roles and features added. The first start
of a Nano Server deployment is literally a couple of seconds, which includes any
specialization. Reboots are equally fast when required. There was originally talk of
less patching and associated reboots during early technical previews, but Microsoft
has shifted to a single cumulative update (CU) once a month for Windows Server
- This change avoids having every organization deploy different combinations of
patches, which made any comprehensive testing matrix impossible for Microsoft and
was the cause of problems hitting customers in production that were not found during
testing. By having a single, cumulative update that contains all updates, a singular
testing platform is provided, as well as greater confidence that any problems will be
discovered in testing and not in production. Because of this single CU, Nano Server
will not have fewer patches; it will simply have one patch, like every other deployment