From the most basic tasks in our day to the products we buy, automation factors in on a massive scale. Automation is a benefit, allowing a range of businesses and industries to achieve greater levels of productivity more quickly and at reduced cost.
In the cloud automation is no different, playing a huge role in how a business can better utilize time and resources to provision infrastructure to support the mission critical applications that generate revenue. However, automation has side effects too. While a process standardization, automation is not immune to design flaws and even human error.
As Keith Allen from Community CA discusses:
“Cloud automation is a foundation for orchestration, but the two terms might be incorrectly interchanged – automation is processes programmed for repeatable tasks while orchestration joins dissimilar automated processes together using workflows.”
One of the top challenges where automation is concerned is not only navigating orchestration vs automation, but understanding how to reconcile a business’ knowledge about cloud with its vision for cloud.
Cloud strategy influences how automation is utilized. If a company doesn’t have a vision that relies on incremental adjustments and is only making decision based on large scale adoption, that organization runs the risk of creating gaps between the scope of automation for the different projects.
However, if that same company is choosing to first iteratively build out an entire private cloud infrastructure, for example, the process of automation is very much tied to the structure of IT delivery the type of cloud supports and can be tweaked over time.
But automation relies on human maintenance to effectively work, and with this human element comes the potential for serious failure. For example, say a company has a large codebase roll out, and they are rolling out a new version of the application code. It’s a 2 gig push to all their servers of which there are several hundred. For this example, Puppet is running across all servers, but to facilitate the code roll out this hypothetical business implemented an internalized version of bit torrent tied to automation. So they would seed a couple of servers with packages and everything would download.
What if someone fat-fingered one of the bit torrent files as they were plugging it into the automation script. If the code is off enough, there’s the possibility for serious security or functionality issues in that code roll out, which because of automation, is being seeded to the entire organization.
Automation is an important part of accessing the value of cloud. But it’s important to know where things can go wrong.
What have been your challenges with automation? Let us know on Twitter @CloudGathering.
By Jake Gardner