How to streamline your IT ops in five steps
Through a mix of automation and smart choices, the Financial Times has established an IT operation that each week helps it push hundreds of improvements to the software and services used by the newspaper group.
So, just how can businesses learn from the FT's success and get its IT operations team working more closely with developers to benefit the bottom line?
Don't reinvent the wheel
There is already a plethora of public cloud services that provide the building blocks for many of the apps and services your business uses.
Before setting out to build a new software tool or service in-house, the FT's director of engineering Rob Shilston recommends first considering whether there is already a service or set of services that your users could take advantage of instead.
"The starting point is an organisation has to think about how much wheel-reinvention it wants to happen," said Shilston.
Even where your business's needs are truly unique, there are probably still tools available that it can make use of, he said.
"In which case should you embrace that [difference] and say 'What's the common tooling that can support the needs to do different stuff all the time?'."
The ease with which a firm's developers can use public cloud services and infrastructure creates a high bar for in-house IT to match, said Gartner analyst Cameron Haight in the report You're Not Doing DevOps If You're Not Focused on the Customer Experience.
"The ability to easily consume public cloud services in a cost-efficient manner creates an expectation of value that DevOps initiatives need to be prepared to address."
Remove unnecessary roadblocks
By not creating unnecessary in-house systems that employees have to navigate, you can make it easier for them to access the services they want.
Shilston gives the example of letting developers deploy applications to the Amazon Web Services (AWS) platform directly, rather than having to use an internal tool created by the FT.
"The change is to say to them: 'Use AWS as a tool, but you also need to be present on our central monitoring system'," he said.
The other advantage of letting developers directly use services on AWS and other cloud platforms, he said, is that the in-house IT team no longer has to play catch-up and add support for the latest cloud platform features to the in-house software deployment tool.
"We said 'If we want our teams to use the new things like [Salesforce's] Heroku or AWS Lambda, then they should be able to when the vendors release it and instead we'll just give them support in how to use that'," said Shilston.
"A good analogy is that previously we were building a road to make it easier for people to cycle along and now we're just providing a few stabilizers," he added.
Small is beautiful
With software being integral to the operation of almost every company today, it's increasingly important for developers to iterate rapidly.
Key to the fast pace of development at the Financial Times, according to Shilston, are microservices.
In microservices architectures, apps are built as a suite of small, semi-autonomous processes that perform specific tasks and use APIs to communicate with each other. Microservices are meant to be easy to use and scalable, and increasingly figure in web, mobile and IoT apps.
The next generation of the FT's website is composed of around four dozen different microservices that tap into the content publishing pipeline.
"It limits the blast radius of a mistake and it limits the boundary of how much knowledge someone needs in order to makes change," said Shilston.
"You can be the expert at how your kitchen drawer works, but you don't need to know how to plumb in the sink."
Everything should be code
Also central to the rapid rate at which the FT can improve core systems is automation of each stage of the pipeline, from development to deployment.
The productivity that allows the team to deploy hundreds of releases each week stems from this automated pipeline, with new code being pushed from Git into AntHill Pro for building and testing, and then into QA, test and finally production environments, all managed using Puppet.
This level of automation is possible because each of these stages can be configured and managed using API calls.
"That is the key message. Everything you're doing needs to be code, the software-defined datacentre and the software-defined network," said Shilston.
"The transformation of having your traditional infrastructure and operations engineers realising they need to write and work with code is a current and future challenge for CIOs."
This release-management process is crucial to address as it "is an activity usually fraught with friction as a consequence of the speed mismatches between development and operations", according to Gartner's Haight.
Move further into the cloud
"I think everyone should be, saying 'The cloud is going to be the future' and the question is whether it's one year, three years or ten years for your organisation," said Shilston.
For "every new thing" built at the FT, on-premise will be considered as the "third or fourth" option, with higher priority given to cloud platforms such as AWS and Salesforce's Heroku.
Shilston expects that, over time, the FT will directly manage less and less infrastructure itself. For example, having already largely switched from managing physical servers to virtual machines on AWS EC2, the FT is now moving away from the hassle of looking after even virtualised infrastructure, by adopting zero-administration services such as AWS Lambda.
"We're definitely using fewer physical servers than we ever were, and I think we will have fewer virtual machines as we start moving more things further up the value chain," said Shilston.
While such a move is not necessarily cheaper, it has definite benefits, he said.
"I believe it will bring greater speed and greater creativity, which will improve diversity of the teams because there's less of the heavy lifting work."