IT Brief New Zealand - Technology news for CIOs & IT decision-makers
Story image
Interview: The finer points of IT automation in the modern enterprise
Tue, 19th May 2020
FYI, this story is more than a year old

From the days of mainframes to today's virtual machines, it takes hindsight to understand that automation was an inevitable result of IT evolution. There are two different types of IT automation: automation of operating systems, and the automation of cloud.

Today most organisations integrate some form of automation in their business - whether it's patch management or DevOps processes, automation offers something for everyone.

We spoke with OSS Group's senior consultant Glen Ogilvie to understand the history of automation, how organisations leverage it, and tips to help organisations understand where automation fits in their business.

The history of automation

The dawn of virtualisation – when many virtual machines run on a physical machine – resulted in the start of configuration management. Tools such as CFEngine started popping up around this time – these could automate configuration across many different machines.

“Automation is used not only to provision, but also to manage servers throughout their entire lifecycle. While some IT management tools concentrate on provision, it's not practical to provision manually when you're talking about thousands of resources. That's where automation started to come in," says Ogilvie.

Now, providers such as Red Hat and OSS Group work directly with their customers - OSS Group's case study provides a practical demonstration of how a local institution used automation for server and patch management.

Automation - cloud

Cloud allows IT admins to define servers and infrastructure as code through APIs. These APIs enable admins to create virtual machines, networks, or almost any cloud use construct. Primitive cloud tools were perfect for organisations that needed to manage only a few IT systems – a few button clicks could create what they needed.

But as organisations grow, they must manage more systems. As more admins use these primitive tools change bits and pieces, no one person knows how the server is configured because there are just too many changes happening. Fragmentation becomes a problem.

In the last five years, Amazon, Microsoft, and other providers have joined the IT automation movement, creating a serverless environment and automation tools built by cloud providers. This has led to a flurry of updates from independent tool developers to ensure that everything can integrate properly.

The market's take on automation

Such an overview of IT automation reflects the problem-solution nature of tools and their development, and Ogilvie says that the market is moving towards automation that can manage an entire system lifecycle from production through to replacement.

“It's becoming more commonplace that nobody logs in or has any manual interaction at all with a particular machine. It provides security because nobody can modify it, and it means organisations can launch more machines with fewer people managing them and more automation. A small team can manage many systems in situations where the systems are all the same – that leads to consistent automation, better reliability, and better security.

Automation also provides the ability to rapidly provision and deprovision systems, which can help businesses scale up and meet demand – whether it occurs in a local environment or in the cloud. It also helps businesses do the opposite: scale down when necessary.

“With great power comes great responsibility. Automation also lets you break things at scale, and it can be expensive,” Ogilvie warns.

“Typically, automation processes involve test environments before production environments. This is so that you write your automation and test it on a single system to make sure it's okay. You then test it on many systems to make sure it isn't broken before you bring it to production specifications.

Automation deployment in the enterprise

“Sadly, automation technologies are not often understood very well by management – they understand that automation can help them, but they haven't necessarily looked at how it works with their business," says Ogilvie.

Automation also exists on a scale – think in terms of no automation, some automation (which is where most businesses sit), and full automation like DevOps pipelines. Automation can be cost-effective and consistent if businesses automate many things that are the same – and if they have the staff with the right skills. You can't automate something if you don't know how it works, says Ogilvie. For example, if you don't know how Google Cloud works, how can you use it to automate?

When businesses move towards automation, it's generally for a new project. New businesses have a bit more flexibility with automation as they don't have established systems yet.

Organisations might see automation as a way to achieve better consistency across IT, but they still have to work with security teams, change management teams, and backup teams. All of those teams need to know how an automation process works.

Not all automation tools are equal – Ogilvie says the most suitable tools match skills that IT teams already have and what they want to be able to do in the future – which is often an unknown. For established enterprises and new businesses, they could be at different ends of the automation scale.

“One of our customers recently had about 700 machines that they were automating across, and a team of two or three people looking after that configuration.  If they needed to change something, they had to raise a change request that says, I'm going to change this across all of the production systems'. The change management team feels a little uncomfortable about that because they used to allow IT teams to make changes to one system at a time," he explains.

Change management and risk are amongst many questions to ask, particularly in terms of cost and failure models. For example, is the cost of unavailability for a period of time less or more than the cost of 100% availability? Ogilvie says it's not always easy to figure out the answers.

“It's better to have the ability to recover from things breaking, and the ability to spin up new things if something else is broken. You can often deploy what you had before on a different platform or region fairly rapidly.

OSS Group built its business around helping organisations tackle issues such as automation, IT management, and much more. To learn more about how OSS Group can support your automation projects, click here.