The Airline Industry's Problem with Absolutely Ancient IT - Wendover Productions
IT failures. Compiler errors. The mere sight of these messages can cause at least a mild annoyance, or even make your entire day disastrous. You probably have experienced these feelings before.
Legacy systems - systems which utilise programs and lines of code that have not been updated for decades - can have disastrous effects on society.
In fact, I know some industries that have still not upgraded their operating systems. For example, I was at the doctor’s office a few months ago, and their computers were still running Windows 7, which reached its end of life more than 3 years ago. This was despite Microsoft’s advice to upgrade, as security updates were no longer given. Many UK hospitals were still running Windows XP in 2017, and the Internal Revenue Service (IRS) in the US still uses COBOL code that was written in the 1960s. (GAO, 2023).
It is quite understandable that some companies still rely on older systems as part of their operations and workflows, but their inaction has led to many disastrous consequences. One particularly recent and notable example of this is the Southwest Airlines meltdown of 2022.
At the end of December last year, Southwest Airlines experienced what was termed “the most catastrophic meltdown in aviation history”. It was initially triggered by the North American winter snowstorm (Storm Elliot), but this led to the collapse of Southwest’s computer system.
Southwest used a software program called SkySolver, developed by General Electric, that allowed them to reroute aircraft if there were issues due to weather on the route, mechanical failures and others. The problem with the system is that it was only able to handle about 300 flight modifications in 20 minutes, and additionally it was not able to analyse crew availability. So, this program started to create nonsensical routings (diverting flights randomly) as the snowstorm caused disruption to many flights. It did not know that air crews would time out, so flight attendants showed up to gates with no planes, pilots to planes with no flight attendants, and many other combinations. This got worse and worse until Southwest decided to perform a hard reset on the system, and switched to manual operation. Ultimately, 71% of Southwest flights were cancelled across the US on Boxing Day, and Southwest posted an estimated total loss of $150 million US dollars for that week alone.
In addition, after all these disruptions, scrutiny turned to the antiquated reservation and ticketing systems of not just Southwest, but other airlines as well.
Have you ever booked a flight, or watched someone else do it? Typically, you would visit the airline’s website, a travel agent, or an online site such as Expedia. They have really nice graphical user interfaces in your web browser, but in fact, it just sends commands to “cryptic” systems. These form the core of the interconnected airline reservations systems, also known as the Global Distribution Systems (GDS).
These were created in the 1960s, when American Airlines wanted to provide a direct link for travel agents to make reservations on its flights. Originally written in COBOL, the system was called Sabre, the first global distribution system. Over time, other airlines joined up on the network, and today, almost all major airlines participate in the major GDS networks of Amadeus, Sabre, and Galileo. These were revolutionary at the time, as it allowed interoperability between airline systems and travel agents. For example, airlines could send bags to connecting flights on other airlines. Over the years, these systems have gained more features, such as the ability to book hotels and car hire with flights.
There are actually two sections to booking a flight. You should have heard about “making a reservation”. In fact, this is only the first part of booking a flight, as it is when you or a travel agent searches for flight routes, dates and times. You might enter your departure and destination cities into a web browser, and then what actually happens is that the website sends a GDS command.
For example, I want to fly from Sydney to Los Angeles round trip for a holiday from the 15th to 29th of June. I enter this into Google Flights:
Looks very nice, doesn’t it? But what Google Flights actually does is send a command to the Amadeus GDS like this:
What it returns is something in the form of this:
Then, the same process is repeated to find the return availability, such as:
So, this is certainly less user-friendly than what we as customers see. In fact, this is the interface some travel agents still use:
After this, we would select a flight we would want to book, and so the second part of the process begins.
This is known as the ticketing process, and once both the flight and class of service (economy, business or first) have been selected, the airline, travel agent, or a site like Expedia will check for the ticket prices, by first selecting the desired flight and fare code SS2Y1 (command to book Qantas flight 11 for one seat in flexible Economy Class), then FXX or FXP to get the fare. For example:
Now when the ticketing is completed, this is shown, and the booking is finally complete, all that is needed to do is for the customer to pay. For example:
Note: This is only an example, and is not from real system output.
This is the interface that many customer service representatives are faced with when trying to rebook stranded passengers at the airport. Depending on the representative, they may have had extensive experience with this command line system, or just be a trainee who only has basic knowledge of it.
For example, if a passenger was stuck and all flights from New York JFK to Los Angeles LAX were sold out, an experienced agent would know to not use automatic availability and instead search segment by segment, as flights may still be available from alternate airports such as BUR (Burbank), also in the Los Angeles area. But, the representative would have to have the skills to reserve and link the segments together. As a result, many passengers were getting stranded at airports when they didn’t have to be.
When the concept of the GDS was introduced back in the 1960s, it was revolutionary for travel agents and the travel industry, as it allowed access to almost all the airlines in the world, and for airlines to sell connecting flights with each other. But, as with any legacy system, it is starting to become less fit for purpose. For example, when Air New Zealand wanted to introduce a new concept called the “SkyCouch”, allowing customers to book a row of economy seats to sleep on long haul flights, it was very difficult to implement in their reservation system, as the GDS didn’t allow partner airlines or travel agents to book multiple seats for one passenger. Thus, customers booking through travel agents didn’t have the opportunity to access this product.
This is not the only instance where the use of outdated GDS systems has caused disruption to the airline industry. In 2017, Delta Airlines had many flights cancelled due to a malfunction in their GDS systems, where they were unable to check passengers in to their flights. British Airways also experienced a similar issue in 2022. There has been progress though. In 2016, the International Air Transportation Association (IATA), the GDS providers and airlines collaborated on a new platform to distribute airline tickets and ancillary products (items such as bag fees, upgrades etc). The CEOs of airlines always promise that improvements will be made to airline systems, but in reality they don’t actually happen, because the airline industry has just settled on a culture of ‘good enough’.
In fact, one employee from Southwest sums this meltdown up: "We're still using, not only IT from the '90s, but also processes [from] when our airline was a tenth of the size," he said. "And it's really just not scaled for an operation that we have today." (NPR, 2022).
So that is what this whole event came down to. It was a combination of corporate mismanagement, inadequate legacy systems, and also a lack of foresight in the amount of capacity that these systems required.
What are possible solutions to these issues that businesses can learn from this example?
I have already touched upon them, but the steps to take in any kind of improvement to a business IT system would be to:
- Consider whether the system is fit for purpose.
- Then look at the capacity and expected demand of it.
- Finally, look at the age of system components, and whether it is getting difficult to integrate older hardware and software into modern workflows.
- And if so, consider investing into infrastructure upgrades proactively instead of waiting until a major or even catastrophic event.
- The important point to take away from all of this is to be proactive instead of reactive.
Just like in safety and risk analysis, it is always better to be proactive in implementing solutions instead of waiting until it is too late and only making changes in response to a major event. This is what the executive management at Southwest Airlines learnt the hard way.
As a result, this should be a good case study for other businesses to examine, so that the lack of investment into IT backbone infrastructure does not lead to dissatisfied customers, public scrutiny, lack of investor confidence and financial loss.
So that’s it from me this time, and I will be back soon with more on IT issues affecting the real world.
Chris Matyszczyk. ZDNET. (2023). How I learned the hard way about Southwest Airlines’ awful technology. [online] Available at: https://www.zdnet.com/article/how-i-learned-the-hard-way-about-southwest-airlines-awful-technology/. [Accessed 6 May 2023].
Wendover Productions. YouTube. (2023). The Airline Industry’s Problem with Absolutely Ancient IT. [online] Available at: https://www.youtube.com/watch?v=1-m_Jjse-cs&ab_channel=WendoverProductions [Accessed 6 May 2023].
Domonoske, C. (2022). 5 things to know about Southwest’s disastrous meltdown. NPR. [online] 30 Dec. Available at: https://www.npr.org/2022/12/30/1146377342/5-things-to-know-about-southwests-disastrous-meltdown. [Accessed 6 May 2023].
Amadeus IT Group SA. (2022). Use the diagnostic tool with itinerary pricing [FXX] - Amadeus Service Hub. [online] Available at: https://servicehub.amadeus.com/c/portal/view-solution/832211/use-the-diagnostic-tool-with-itinerary-pricing-fxx- [Accessed 6 May 2023].
Alavi, J. Flyingway. (n.d). The Complete Amadeus Manual. [online] Available at: https://air.flyingway.com/books/amadeus/Amadeus_Guide.pdf [Accessed 6 May 2023].
EasyPNR Blog. (2018). Will ever GDS systems be replaced? [online] Available at: https://www.easypnr.com/blog/will-ever-gds-system-be-replaced/ [Accessed 7 May 2023].
Office, U.S.G.A. (2023). Outdated and Old IT Systems Slow Government and Put Taxpayers at Risk | U.S. GAO. [online] www.gao.gov. Available at: https://www.gao.gov/blog/outdated-and-old-it-systems-slow-government-and-put-taxpayers-risk [Accessed 9 May 2023].