The topic of automation and orchestration comes up in just about every conversation we have with customers. Recently, the idea of configuration management has been brought up about as often, and it is only natural to consider both concepts together. But does that mean automation requires configuration management or vice versa? Not necessarily. It turns out that you can start with either and be successful on your journey to automation nirvana.
Getting started with configuration management
You already know how to make rock solid templates and clone virtual machines from them. What most organizations don’t have a firm hold on is the idea of configuration drift away from a known good state as the weeks and months go by. Someone changes a registry setting which violates a security policy or changes a memory setting on a SQL database. The question is: How do you know when this happens? The answer is typically: You don’t.
Configuration management is one easy answer to this problem (and a lot of other use cases). Take a solution such as Puppet. All of my SQL Servers have a standardized SQL Server manifest that represents a pristine SQL Server configuration. At some regular interval every server in the group is checked against that manifest for the appropriate settings. If something drifts, it is automatically logged and brought back into a known good state.
Getting started with configuration management requires a very approachable amount of work. Prioritize the most important aspect of configuration in your environment and start there. Perhaps it is SQL Servers, or IIS Servers, or maybe just ensuring basic OS settings are properly configured. As the solution grows you will have more time to design sophisticated automation.
Getting started with automation and orchestration
Many tasks can be automated, from creating storage volumes to finding a free IP address. By creating automation building blocks, engineers can create orchestrated workflows that produce a result greater than the sum of its parts. For example, a “self-healing data center” where compute capacity is automatically replaced when failures occur.
Once again, a start small approach is best. Many organizations simply begin by automating the task of cloning a VM from template. From there the possibilities open up to include pre- and post-clone workflows such as finding an IP address, installing database software, and even automatically instantiating an application. Configuration management is a great adjunct here. Using our previous example of SQL Server, a new VM may be automatically recognized as a SQL Server and moved in to the corresponding group (by some criteria such as name or OU location). Once in that group it will automatically be checked against the SQL Server configuration and found to be lacking SQL Server. The configuration management solution can then install SQL Server and configure it per the known good configuration.
In this example we started with simple automation and orchestration. The process did not require configuration management, however we can see a later stage where it has been added to our benefit. This is the iterative process we suggest and evangelize to all of our clients.
It doesn’t matter whether you begin your automation journey with configuration management or orchestration, because giving consideration to either of these topics is already a win. So many people tell me, “I’m too busy dealing with project work and closing tickets to design automation workflows.” The fact is that starting small will take little time and begin a very small, but steady, ripple effect. After the first task is automated you may have time to automate one or two more and so on. Pretty soon you have half of your time freed up by automation to work on more automation. Start wherever you wish – just make sure to prioritize and have a plan. We can help with that plan. Schedule some time with us in AHEAD Lab and Briefing Center.