BACKGROUND IMAGE: iSTOCK/GETTY IMAGES
When organizations automate IT tasks, the ability to do one thing over and over again is a starting point -- not the finish line. This technology is getting smarter and more flexible.
Computers repeat a task with the same output every time, but it's a human who tells them what to do. This was shown in spectacularly negative light when Amazon Web Services suffered an outage in February 2017. A systems administrator accidently took down more servers than intended, starting a domino effect that then brought a large part of AWS' infrastructure offline.
The problem started when a human did something wrong, but a lack of intelligent IT automation technology exacerbated it. The platform did not recognize an issue and replicated it as programmed to. If it can happen to Amazon, it can happen to you. One of the biggest IT automation trends is technology with checks and balances in place to prevent bad decisions from spreading out of control.
Plan automation well
IT teams can program a system to automate a single task for a single iteration -- but why would they? It starts with handcrafting the script or code to carry out the task anyway; automation accomplishes nothing more than what any sys admin pressing Go would. IT systems automation shows its strength where a task is carried out repeatedly, either at different times against a single system or against multiple ones.
Systems patches and updates provide a prime example of IT automation benefits. An OS patch is brought into a system for initial test. The sys admin puts the patch onto a test system manually, looking for any bad consequences before it rolls out across the live environment.
From the start, IT should use the same tools in test that will be used in operations. This helps prevent the sys admin and IT automation platform from diverging -- if the sys admin takes the same actions that then become a scripted flow, it prevents unforeseen problems.
In addition, this uniformity lets the admin proof any further actions that the IT automation platform will take. For example, the tool deploys a patch, and the patch does not work, which requires a corrective action that can be programmed in as well.
Intelligent IT automation technology
Intelligence tops IT automation trends now. Dumb automation, as used in many basic script systems, can bring everything down.
Remediation should be a capability in any IT automation platform. It can take the form of rollback, wherein the platform identifies a problem that it cannot deal with directly and therefore returns the whole system or any specific parts of it back to a working, known state. The rollback process then alerts people, who use logs and other data to identify what caused the problem and take actions to fix it.
Remediation can alternatively occur when the automation tool identifies why a system did not take the patch and changes it so that it can. This kind of remediation should be preemptive: An IT automation platform should examine all the target systems before attempting to roll out the patch and single out which ones cannot take it. The system must then take direct remediation or raise an alert. In some circumstances, the fix requires a hardware change or complete replacement that is outside of IT automation technology's capabilities.
Here to help
Systems such as Chef, Puppet, Automic as part of CA Technologies and HashiCorp Terraform offer good levels of automation support for DevOps, with the latter two providing the in-depth intelligence required to adequately manage such complex environments. Systems management tools, such as those offered by BMC Software and CA Technologies, have adapted to meet the needs of complex environments. Some data center infrastructure management tools, such as Nlyte and Future Facilities, also have good capabilities in the space.
Another IT automation trend is reusability as a platform feature. Whether a step or task can be reused depends on the path to automation that an organization chooses. For example, if a script provisions one certain workload on N number of virtual servers, it only saves manual work when the user wants that specific workload to spin up. However, if the tool creates a workload container, the script can say to provision workload A on N virtual servers. The same script also works with workload B, C, D and so on.
The organization can change the contents of workload A, B and others as needed and adjust the way in which the specified workload is provisioned. The abstraction layer between the provisioning and package automations creates an object hierarchy that gives the platform user greater flexibility. The sys admin selects script A from the workload creation side and wraps it in script B from the provisioning side to achieve a desired result.
Organizations can automate IT tasks beyond the OS level. The same approach applies to application stacks, containers, end-point devices and firmware -- any system that currently requires a highly paid, yet fallible person to administer.
IT automation has entered the realm of necessity. It's impossible to provide a steady and stable platform that combines physical, virtual and cloud resources across a range of public, private and hybrid models if you don't automate IT tasks.
IT teams must investigate which automation technologies make sense now -- before a lack of repeatability becomes a major concern about IT performance.
Automation is happening beyond the IT level, across the whole organization. Business process automation should also be a strong point for IT, which must provide the necessary technology for smooth and effective BPA use. As the internet of things weaves its way into the organization, many thousands of devices will require ongoing security and maintenance. Expect to invest in extra automation capabilities to serve them.