CIO interview: Rorie Devine, CTO, Hailo

Replatforming the IT stack

The response to that objective was the complete replatforming of Hailo’s IT setup, which went live two months ago to create an open-source, cloud-based, API-fronted, microservice-based architecture.

“Before, we had a traditional stack – obviously, we had our iOS and Android apps, PHP and Java, APIs, and engines and databases, which was good. But we grew very quickly and thought that to do really well everywhere in the world, we needed the absolute maximum scalable architecture we could possibly have, and that’s what we designed, and that is what we have delivered,” Devine says.

Hailo is now 100% open source and based on technologies such as the Go programming language, as well as Google's XML equivalent Protocol Buffers, message-oriented middleware RabbitMQ, real-time message processing system NSQ and database management system Cassandra. Hosting is provided by Amazon Web Services (AWS).

“It’s a much more scalable architecture, making it much quicker to get stuff done,” says Devine.

“Plus, culturally we wanted to be part of the open-source community, using and building shared libraries and feed that back to the system.”

The first city to go live on the new open-source stack was Atlanta, followed by Manchester and Boston. A few cities in Asia will go live next in the coming weeks and Devine expects all the cities in which Hailo operates will be migrated by year end.

Ongoing improvements

One of the first decisions was to run the replatforming in an agile way, with a two-week “scrum and sprint” cycle.

“This was unusual for a replatform, because obviously a platform is, if anything, a prime candidate for a waterfall project. So what I was saying was that to fit our culture and the speed in which we wanted to achieve this was to do it as an agile, sprint process,” Devine says.

“So we kicked the ball around in June 2013, and said, ‘Right, by the end of July we want to be taking real points into our new platform’, meaning that every five seconds, the Hailo driver sends a GPS point back to the base to say where he is so we can track where the drivers are; we can track journeys, and we get a lot of data,” he adds.

The transaction volume has grown massively and we didn’t want to hit a brick wall, so we built a really horizontally scalable architecture that is also global, allowing us to write a feature and have it used all around the world if we want to

Rorie Devine, CTO, Hailo

“We worked amazingly hard and achieved that goal.”

Now the new platform is in place, one of the main areas of focus will be scalability.

“When we created the replatforming initiative, we wanted to ensure there were absolutely no points of failure, so it’s as close to 100% reliability as you can possibly get – and so you need architecture to do that,” says Devine.

“We needed an architecture that could scale. The transaction volume has grown massively and we didn’t want to hit a brick wall, so we built a really horizontally scalable architecture that is also global, allowing us to write a feature and have it used all around the world if we want to. And that’s what we did,” he adds.

“I think that in future, we could be talking about scalability and in AWS our services autoscale, so they will automatically add CPU power to our capability based on demand, so we are not going to hit a brick wall.” 

However, the change process is ongoing, says Devine. “A platform is definitely not something you do and walk away from. It’s now live in three cities and it will be constantly improved; we will be always interacting to make it quicker, better, faster and cheaper.”

Running on cloud

According to Devine, Hailo has a good relationship with its key supplier AWS and feels that both parties are in a “strategic partnership”. He rules out any concerns around a decline in the quality of service prompted by the cloud provider’s exponential growth.

“AWS is growing incredibly quickly, it’s hard to deliver consistent service when you are in a hyper-growth environment, I’ve been there, I’ve tried it. But I think they caught up again, to be fair. I am not giving them an easy time, but I think that they spent a lot of time in the past six months just making sure that they have the service levels and the customer relationships to where they should be,” says Devine.

“What I like about the Amazon culture is that it is genuinely customer-focused; it’s right in the heart of what it set itself up to, and it’s in its values and principles. Those are things that are really good when you are dealing with a supplier and it is clearly not just a mission statement – it’s something they actually do,” he adds.

“That’s how I feel, [but] we are not holding on to them, we are platform and supplier neutral, so we would go to the place that we think is the best for the Hailo business. One of the reasons why we are totally open source is that we don’t get to be a victim of the locked-in model, which I experienced earlier in my career. ”

Hailo has a high volume of transactions that involve a great deal of sensitive data – so having data hosted externally would typically pose concerns to most decision makers in Devine’s position. But the IT chief takes a pragmatic view on cloud.

“Private clouds are meaningless. They are made up to keep people happy. If you run a datacentre and you are feeling a bit untrendy, you can call it a private cloud. If you think you are going to get heat from your management team, you can call it a private cloud even though it is public and serves a lot of people,” says Devine.

That does not mean Devine is not concerned about data protection. According to the CTO, Hailo took a conscious decision not to store credit card details.

“Obviously, we take in the details, encrypt them, pass them on to the banking party but we do not store that data, so it can’t be stolen from us,” he says.

“People talk about cloud as if it is inherently less secure, but when you have a company the size of Amazon, with the power at its disposal to ensure security, you are going to be a lot more secure than the one or two security guys that you can employ in your business,” he adds.

“These companies have pretty good specialists ensuring that people can’t just get one pinchat Amazon and then steal the data. There were situations where they notified us about things, which shows they are proactively examining the security profile of their customers to ensure their security and notifying you if things can be improved, as it has happened to us.”

Organisational change

Devine says the timescale of the change programme was ambitions and the intention was to maximise delivery, so some of the consultation and collaboration had to be cut short to achieve that.

“We are talking about a very high-performing team who delivered an amazing platform in an extremely short period of time, but as someone who has been around the block for as much as I have, looking back, I would have attempted to have done that less divisively,” Devine says.

Within a year, Hailo’s IT department has doubled in size and now employs 100 people. However, the team behind the platform transformation consisted of just 12 members of staff.

“The team grew in proportion to the business growth, less so actually, but I don’t want to run a bigger IT team at Hailo. I would be happy to run a 100-person team forever; our challenge is to keep it agile, small, lean and quick. Having said that, we only hire people when we really have to,” he says.

However, the IT team at Devine’s old Betfair job grew massively even though the intentions were very similar. How is Hailo different, given that it is looking to hire IT professionals?

Private clouds are meaningless. They are made up to keep people happy. If you run a datacentre and you are feeling a bit untrendy, you can call it a private cloud

Rorie Devine, Hailo

“The difference is that [Devine’s time at Betfair] was almost 10 years ago, and the tech capability and the ability to scale is so much better now than it was. A lot of companies such as Google, Facebook and Amazon have faced this challenge of scaling effectively and not adding thousands of people,” he says.

“Also the Betfair model was not like any model – a lot of the people requirements came from the fact that we had to have local IT and local infrastructure, so we ended up having operations in places for tax or licensing reasons rather than pure technology reasons.”

In a self-deprecating way, Devine admits he has changed from the type of IT leader he used to be back then.

“I learned I wasn’t very good at it 10 years ago, and I’ve learned a hell of a lot during that time. What I try to do now is to be more collaborative than I was in the past to really engage the talent in the team, and the decision making, ideas and creativity, to really use all the talent within the team as much as I can,” he says.

“I was used to being a leader and ultimately, you don’t have too long before the execution; you just want to get stuff done. But I hope that the experience of working with me now is where you feel like you are much more a part of a team than before; I am much less directive inside teams now.”

Saying that “Spotify and Netflix were once a baby that looked like Hailo", Devine expects that under his leadership, the IT department at the company will remain agile, and experiment with leading-edge technology in open source and consumer tech such as Google Glass, but most importantly, will be ready to support growth.

“I think you might be talking to me in a year’s time as a CTO of a business that will be up there with your Amazons, your Facebooks, your Googles. I think it will be that big and that important. We are not there yet but we have great potential.”