Richard Haigh, Director of Technology (SRE) at Just Eat and Bennie Johnston, Head of Technology at Just Eat explains how DevOps at scale helped create “the world’s greatest food community” at AppDynamics Global Tour 2018
Just Eat have “transformed the restaurant marketplace, providing customers with an unparalleled choice – whatever, whenever and wherever they want to eat – all via software”. IT has been central to their success, so in Haigh and Johnston’s speech, they explained how they got “DevOps right”, as well as best practices for building site reliability engineering teams.
The food company has more than 500 technical workers in the UK, Ukraine, Australia and Canada; more than 30 teams; around 450 services; receives 2,700 orders per minute; 1,500 Amazon Web Service (AWS) instances in production; 1.6million metrics per minute; 1.5tb logins per day; over 500 releases per week; a revenue growth of 45% last year; and a £5billion market cap.
According to Haigh, keeping up with the pace of change in technology is relentless, which is why DevOps is so important to them – as well as “playing around with robots to deliver Just Eat’s grub”.
“We innovate to stay ahead of the game through tooling, to ensure that we are constantly delivering the best products and tools,” said Haigh to the conference attendees.
“We need our teams to deliver infrastructure that distributes tens of thousands of releases per day – without any false positives!”
To help achieve this, they run a daily standup first thing in the morning, weekly risk meetings and get all technology together once a month to discuss issues that have recently come up as well as new, innovative ideas to help transform the company.
5 key SRE principles
They both believe it’s important to follow the 5 key principles below in order to operate Just Eat’s site engineering:
- Relentlessly protect site availability
- Enable change to be delivered fast, but with quality
- Optimise the use of infrastructure and resources
- Innovate to stay ahead
- Foster the right culture at Just Eat
“If you run a DevOps team at Just Eat and want to try new tools we will support you and make it easier for others to adopt,” he added. “All of our teams are responsible for code and production – providing deployments and monitoring tools.”
Nevertheless, according to Johnston, there was a lot of issues when trying to scale DevOps at Just Eat.
“To help us, we brought in an operations team to monitor what’s going on, while looking at what other big companies are doing – it’s good to have your eyes on the platform,” said Haigh.
The biggest challenge
Through both of their times at Just Eat, they have helped create more and more DevOps teams, although, according to them, their biggest challenge has been from the operations side.
Johnston continued: “We managed to get where we are with help from a tool that we created called Score Card, which shows how well our teams are engineering through checklists and individual scores.
“Another tool that was useful was AppDynamics as it lets us see what’s really going on in our systems – before this, we had monitoring as a service which came from an individual service but this showed little visibility from the operations team.
“It’s important to look at patterns to ensure we are where we should be – highlighting good behaviours and team patterns.”
For Just Eat’s future, they revealed that they want artificial intelligence (AI) to help towards the company’s business transactions and that their dream is for bots to identify and understand implications of what needs to be done.
Johnston also noted that “AI is the start of developing products more and more efficiently!”
Written by Leah Alger