Puppet for vSphere Admins
Over the last four weeks, we looked at Auto Deploy, which is automation for our VMhost provisioning process. Next up, we’re going to look at Puppet, a tool to automate our VM and Guest OS provisioning.
Recently, I have been working on deploying Puppet by Puppet Labs in our work environment. Puppet is a provisioning and configuration management system. It has been made famous by its ability to simplify cloud management for those running at scale, such as the Obama for America campaign that leveraged AWS and Puppet. Manually provisioning and configuring nodes scales linearly, or worse – 5,000 nodes requires at least 1,000 times the resources as 5 nodes. Automation with a tool like Puppet scales much more gracefully. Managing 5,000 nodes is only incrementally more difficult than managing 5, and growing to 50,000 or shrinking to 500 is just as easy. There are a number of other similar products you might be interested in – Chef, Ansible, and Salt to name a few.
I am interested in Puppet for two primary reasons. First, it has a lot of mindshare and a friendly community. You can easily find numerous blogs addressing common problems, there’s an active irc channel for problems you can’t solve with the help of a search engine, there’s a gigantic public module repository (called Puppet Forge), and you’ll find many candidates who already know Puppet as you grow your team. Second, VMware has invested $30M in Puppet Labs. The increased interaction and development has already resulted in Puppet adding some VMware cloud provisioning features and should ensure those features mature. This should help round out their Software Defined DataCenter (SDDC) efforts.