We will go into DSC architecture in much greater detail in the next chapter, but it is useful to show a quick overview of how all the concepts we just covered fit together. DSC has several steps that can be bucketed together into three large phases.
As we mentioned earlier, DSC can manage more platforms than just Windows, but for the purposes of this book, we will stick to Windows configurations. The DSC Resources for Linux and other platforms are in the early stages, so there will be significant changes as these implementations mature; it would be too hard to account for these changes in the space we have in this book.
You begin with DSC by writing a configuration script in PowerShell. The script itself doesn't actually do anything. You can run the script interactively all you want; it won't change a thing. Since the configuration script is the DSL we were talking about earlier, it's only a list of things to do, not the things that actually execute the list. Because there can only be one MOF per target host, and each configuration script is translated to a MOF, this means there is usually only one configuration script you write, which handles all the variances in your environment. This sounds like it will get complicated and be difficult to manage quickly, but there are DSC patterns to follow to manage this. We will cover these in Chapter 3, DSC Configuration Files.
The next step is to translate the configuration script into a MOF file. The translation, or compiling, happens only once—when you deploy the MOF file to the target node or to the DSC Pull Server. The configuration script is often kept in a version control system and only compiles and deploys the MOF file when the configuration script changes.
The next step is to get it over to the target computer. The deployment of the MOF happens in two ways: push and pull. A push method is when you execute the Start-DSCConfiguration Cmdlet, which compiles the MOF and copies over to the target system. The pull method involves putting the MOF file on the DSC Pull Server, which handles distributing it to all the target hosts.
Whether an MOF file was pushed (using Start-DSCConfiguration) or pulled (using a DSC Pull Server), the next step is the actual execution of the MOF file. If pushed, the execution happens interactively or in a PowerShell job, depending on how you called the Cmdlet. If pulled, the Local Configuration Manager (LCM) schedules and executes the MOF file without user input or oversight.
LCM is part of the DSC system installed on the target node and is responsible for receiving, coordinating, and executing configurations on target nodes. LCM itself is configurable using DSC and is flexible enough to allow multiple types of deployments.
The phases described earlier will be covered in much more detail in the coming chapters, so do not worry if some of it does not make sense.