Gatwick’s ambition for Splunk-based analytics reaches new heights


Off the back of several successful projects in real-time operations, Gatwick Airport’s head of business systems Chris Howell is now looking to explore predictive analytics.

Back in October 2016, Gatwick Airport’s hopes of winning backing for a second runway were dashed when the UK government decided instead to give the go-ahead for a third runway at rival Heathrow.

That’s not to say, however, that expansion is off the cards at a transport hub that its bosses like to call, ‘the world’s busiest single-runway airport’. In January, they appointed a new chief planning officer Tim Norwood to help deliver a £1.2 billion investment programme at Gatwick.

Demand among air travellers is on the rise, too: 2016 passenger numbers at Gatwick were up 7%  year on year, to reach 43 million, and the airport has just experienced its busiest-ever January, up 12% on January 2016.

Back in October 2016, diginomica reported that Gatwick Airport had exceeded its target of processing 95% of passengers through security within 5 minutes, thanks in part of its use of analytics technology from Splunk.

More recently, we’ve caught up with Chris Howell, head of business systems at Gatwick, to hear how that goal was achieved and to learn more about his plans to take Gatwick’s Splunk use to the next level – deploying it for predictive analytics, as well as insight into real-time operations.

Howell’s story starts with the e-gate readers through which passengers pass on their way to security. Today, in the case of international passengers, he says, it takes these e-gates around 3 seconds to scan a boarding pass, validate the flight that the passenger is aiming to catch and open the gates. For domestic travellers, it’s more like 8 seconds, because there’s additional biometric scanning of the iris and face.

But two-and-a-half years ago, Gatwick was experiencing significant problems – and much longer delays – with that e-gate solution, he says.

I needed to understand what was going on across the various pieces of distributed software that underpin the system, in order to do some analysis and fine-turning. It was a good old-fashioned IT operations problem of making this process more consistent and faster, so that passenger experience wasn’t impeded.

Given our rate of growth in passenger numbers that we needed to get through those e-gates every hour on a busy day, the business case for making sure that they weren’t a choke point was really clear. At that point, I went and got Splunk in order to pull together all the bits of log file data and everything else across the applications and integration layers to help solve the problem. It got the job done in a way that really helped the business.

That prompted Howell to go looking for other potential ‘choke points’ in the airport where better insight could mean greater efficiency. He found it just beyond the e-gates, where passengers file into in security lanes and where around 750 trays, containing their belongings, need to be processed each hour:

On a screen facing the security teams working on that, there’s now a ‘Know Your Numbers’ screen, with data from Splunk that tells them whether they’re on target, beating 750 trays per hour, and also lets them know how many trays per passenger were processed in the preceding 5, 10 or 15 minutes.

On the runway

Out on the airfield, too, where around 55 aircraft take off and land every hour, a great deal of data is pulled together in Splunk from the local air traffic control (ATC) tower, the national ATC centre and European ATC provider Eurocontrol, as well as airlines, ground handlers and the airport itself. This is presented to various teams in the form of dashboards to show how they’re performing in on-time performance and also turn performance – the time it takes for passengers to disembark an aircraft and that aircraft to be cleaned and reloaded with passengers for the next trip. Says Howell:

It’s like treating airfield staff like they’re a Formula One crew, showing them how they’re contributing to overall airport performance. Everyone having that data and sharing it with others enables them to collaborate, rather than engage in a ‘he said, she said’ routine of blaming others when things go wrong.

And in times of disruption in particular, such as the heavy fog we saw towards the end of 2016, it allows them to focus on getting necessary work done to offset delays as far as possible, rather than spending time trying to figure out from spreadsheets and so on how they’re doing against targets.

These kinds of insights proved particularly valuable during a series of airline moves that Gatwick completed in January this year, which saw British Airways move to Gatwick’s South Terminal, Virgin Atlantic to the North Terminal, and Easyjet, which previously operated out of both, consolidate its operations in the North Terminal.

Because Howell had previously built a Splunk dashboard for the head of car parks Gary Wallace, he and his team were able to quickly identify and avert potential problems for airline staff looking to leave their vehicles in newly assigned staff car parks:

People often don’t realize how much data they have in the systems they’ve got, but once you bubble up that information to operational teams, they can achieve some real successes. I know it sounds like I’m getting up on my soapbox, but I really believe that there’s a lack of understanding that the data held in log files is business data – and that, as a technology function, you have the keys and the access to a really rich source of critical business information. The trick from there is making it available to the people who can take action on any problems, which is what Splunk helps us do.

This experience has left Howell convinced that Gatwick’s car parking facilities are a natural target for more predictive analytics, along with its airfield operations. So that’s where he’s planning to dig deeper into Splunk:

Before I joined Gatwick, managers knew what had happened last week. Today, they know what’s going on now. What I now want to do for our operations community is help them understand what demand is likely to be on parking spaces, based on previous patterns and other factors.

By this, he means not only the spaces in car parks, but also the high-value spaces for aircraft, out on the airfield, where they connect by piers and air bridges to the main terminals – and in both cases, he plans to bring in other data, such as weather information, in order to help staff make accurate guesses on what spaces are likely to be free, when and for how long. Says Howell:

The reason I chose Splunk at Gatwick was in part based on lessons I learned in multichannel retail before I joined the airport. There, I saw its strength in parsing Internet log files underpinning e-commerce websites bringing in a lot of revenue. Every transaction meant data was written to a log file, to a database or to a systems event viewer as part of an operating system. What Splunk does really well is bring a lot of that information together from all of those different sources really easily.

I’m not a techie. I’m a business analyst and consultant. I’m here to translate, but even I’ve been able to successfully import data from a log file into Splunk with no trouble at all. What’s also key is that the performance overhead of Splunk on critical systems is absolutely minimal, which matters a great deal in a time-pressured environment like an airport.

Image credit - Gatwick Airport