Manchester Airport: A risk-based flight path to cloud migration
The migration from legacy systems to cloud based services is not a smooth journey for an airport – as Graham Pollitt, IT infrastructure director at MAG Airports Group attests.
“Airports are very ‘on prem’ kinds of places – we have planes taking off and landing – that doesn’t happen remotely!” he says.
The group – which owns and runs Manchester, East Midlands and London Stansted is also in operation 24/7 with little downtime to carry out upgrades or refreshes.
Additionally, as part of the UK’s national infrastructure it is also a high profile, high security environment. Pollitt is all too aware that “queues make news” and if there’s a malfunction, “the whole world gets to know about it.”
He recalls the time that Gatwick moved all of its flights over to the cloud and a digger went through its cables: “They should have been resilient but weren’t, chaos ensued, and it make front page news.”
Consequently, MAG has adopted what Pollitt describes as “a balanced risk posture” to doing things in the cloud.
“Because we all know that as much as you have got resiliency and dual network feeds and everything – as we saw with Gatwick and the air transport data firm SITA [which suffered a data breach following a cyber-attack last year] – networks do go down.”
According to Pollitt, the airport’s business is fundamentally split into two different parts.
“You have the operational running of the airport – where you’ve got check in, baggage, security, car park barriers – and that’s more like a manufacturing type of environment.
“Then the other half of the business is running the company: the finance, the HR, the service desk as well as Office365 and email.”
MAG’s approach has been to move anything not to do with the passenger journey out to cloud-based subscriptions. “It’s been about finding the sweet spots to move over – the service desk, the finance – that’s a no brainer,” says Pollitt.
The migration of these services to cloud started in 2019 at Manchester Airport as part of a £1 billion digital transformation investment – with support from technology partners Keysource.
It included new fully resilient high availability data centres on-premises as part of a hybrid cloud environment.
By the time Covid hit, the upgrade meant that all the airport’s back-office staff were already on SAAS-based apps such as Maximo, ServiceNow and Office 365 “which allowed us to facilitate the immediate working from home,” he adds.
The drop off in flights during the pandemic gave the IT department some much-needed breathing room to focus on the core networks of both Manchester and Stansted, Pollitt adds, but these refreshes were done on a skeleton staff given that a “significant” proportion of MAG’s IT team were on furlough, as airport revenue plunged.
“It would have been a perfect scenario to do everything – but we couldn’t afford it. So, we had to maximise the opportunities to do tech refreshes things that were more complicated.”
Simple flight path
According to Pollitt, MAG’s primary consideration has been around simplicity, ease of use and ease of migration. But what about the physical assets involved in the operational running of the airport and moving car parks, Xray machines and CCTV footage to the cloud?
Because airports generate a huge amount of CCTV across the estate, with lots of high-quality video recording, this generates significant demands on the network and compute estate. MAG concluded that it didn’t make sense to offload this type of service into the cloud at this moment, when they can keep it on-prem.
“Other things like the hand luggage X-ray machines which deal with high-res images just make appalling use cases to throw into the cloud because you are transferring massive amounts of data. And things like car park barriers – that’s a physical thing…” Pollitt says.
“The question in the short term – is it a good thing to throw into the cloud? Do we want to throw it in to the cloud? Probably the answer at the moment is we have no intention of throwing that into the cloud, so we are adopting a hybrid approach. We have a bunch of stuff on prem and will stay on prem.”
In the midterm, Pollitt adds that there are areas where MAG will utilise tech refreshes to “design systems in a new way where potentially they are cloud capable.”
The transformation of flight information display screens is a case in point. Until a couple of years ago the airport used greenscreen displays to tell passengers what gate to go to.
These were switched for cloud-based content management systems which enabled technicolour screens that display airline logos and are capable of broadcasting other information, such as adverts for airport shops and the weather in destination countries.
Pollitt adds that at all three airports these displays are backed up on prem with local caching servers so that all the flight displays can continue for a couple of hours, should the networks go down.
OPEX VS CAPEX
One of the challenges the IT manager now faces as services move over to cloud is the balancing the technical debt of legacy kit with the need to innovate to drive the business forward.
He maintains its still easier for him to raise the money for capital expenses – such as buying servers or develop systems – than it is to justify the typical operational expenses involved in cloud computing, such as software licences and storage costs.
“I can spend CAPEX developing systems but the moment that gets written into a SAAS or OPEX cost our bean counters loath and detest me for it. It’s another business obstacle that you must come across in the organisation.
“I try and bring it back to the fact that the whole world is moving towards subscription services. You buy services for you home in that way – with Netflix and Amazon – try asking them for a perpetual licence! But it’s a big challenge for us.”
Pollitt also acknowledges that the procurement process for cloud services also doesn’t match the old model of governance.
“Sometimes the developers can literally say ‘Can spin up three more instances of extra storage?’ You click one button and then suddenly you get a big bill at the end of the month from AWS or Azure. And you must question whether you are consuming all of it.”
He adds that MAG is introducing a whole series of business and governance processes around measuring consumption in a cloud environment.
“It involves looking at those instances and questioning whether we need them all. If you are running a project, and for whatever reason that project is put on hold for 3 months, then you need to be able to spin everything down and make sure that happens. After all, that’s the whole point of being in a cloud environment.”
“It’s the digital equivalent of turning all the lights off when you leave a room. Or looking at your bank account and realising that all those subscription services with a 30-day free trial period you’ve taken out have started charging you for using them.”
Another strategy useful for working in a hybrid environment, and one that Pollitt claims helps staff deal with the mind-set change needed for cloud – has been the introduction of multi-disciplinary product groups responsible for the continual ownership of and running of various functions around the airport.
There may be one around spinning up a separate network for check in, for instance, comprising of network engineers, server engineers, storage engineers and telecommunication engineers.
“It’s about bringing people much more into matrix teams and running those products and continually owning the products and being responsible for their lifecycle, not just implementing them and then throwing it over the wall to IT support,” he says.
He adds that the flight display systems are a good example of this approach.
“They started by saying: let’s just get the raw flight information into the cloud; then we can start getting feeds from other systems and improve the content management aspect of it; then it looked at how to hook in with retail. All those gradual increments.”
“And because you have these product teams you are then able to develop the skill sets of the people in those product areas. They become part of that project or product development that’s going on. We’re trying to move the whole organisation in that direction,” he adds.
Dividing things in product teams also shares the burden of what to do with any legacy tech, he adds, with each team making an informed decision on the best time to “switch off the old stuff”.
“You might have a new car park system to manage bookings by introducing plate readers. If we introduce a new system, it might make sense to migrate all the other car parks over to that system,” he says.
“We might do it gradually, but we get rid of the old systems as we go so, we don’t have five systems in operation – meaning that burden of support is much lower which frees up people’s time to work on further developments and innovation.”
Subscribe to our Editor's weekly newsletter