DevOps: Adapt or Die
DevOps: Adapt or Die
Join the DZone community and get the full member experience.Join For Free
Learn more about how CareerBuilder was able to resolve customer issues 5x faster by using Scalyr, the fastest log management tool on the market.
First Chris Siebenmann, a well known blogger and "herder of Unix", wrote in response to Julian Dunn's " Chef, devops, and the death of system administration" (and the subsequent update on his views)
His basic argument was that what many people today perceive as sysadmin work is "scutwork" - connecting cables, racking servers, restoring files for people, installing OSes, etc. His response to the death of system administration was "good". Less boring grunt work and more creative contributions to the organization's prosperity.
Then Phil Hollenback hollered back. :)
"Like all dinosaurs, we have a choice: adapt or die," agreed Hollenback.
The answer is DevOps. You've got to start automating. You've got to get on the cloud. You've got to make that developer and operations communication start happening…
Let's for the sake of your career and sanity assume then that you aren't going to treat those virutal servers as precious and unique snowflakes. That means you are going to need configuration management and automation. Hey look, those are a couple of the core devops methodologies. And hey, what about the fact that you're now working with automation? That's DevOps again. Hm, I'm seeing a lot of DevOps in your path.
-- Phil Hollenback
One commenter on Phil's post was Chris Westin, a software engineer at 10gen who has also worked at Yahoo, Netflix, and Oracle (basically a programming badass). He went even further saying your large scale applications need to be "fault-tolerant, and/or self-healing." This means that sysadmin skills need to be applied to development and the application needs to be designed alongside the infrastructure.
Finally, the blogging came full circle when Chris Siebenmann responded to Phil Hollenback. There was a lot of argument between people who say "There is never a valid case for not applying config management to servers." and the people who believe that there are times where you only have a few virtualized servers and services, and it would be more work to learn and setup Chef or Puppet than it would be to just keep them running by hand.
With the arrival of PaaS providers like Heroku, Siebenmann says that the only thing remaining that may resemble an operations group will be monitors who keep an eye on the production environment performance and regulate code that is moving from development to production.
He also envisions a future where programs will be shipped with their own Chef or Puppet recipes for various standard setups so that developers don't need any special training to go from writing the app to installing it in the environment they want.
It's a very cool time for developers, and it can be a great time for sysadmins too if they embrace the changes that are coming. For many sysadmins like Siebenmann, DevOps-like practices have been in place for years. Which is good.
Opinions expressed by DZone contributors are their own.