Airline IT strategy
At Gulf Air we’ve been working to ease the deployment of business intelligence, big data, cloud and other technologies to help drive our business and enhance performance and efficiencies.
One striking example is our sharp focus on route profitability and technologies that will enable us to achieve excellence in this area.
Route profitability is critical for any airline. It provides a wealth of information to strategically tune the airline’s network, so you can report on the revenue earned and costs incurred from each route.
In aiming to deliver route profitability excellence, we’ve encountered some challenging times. But it’s been a rewarding experience and we feel it’s certainly placed us in a strong industry and technologically leading position in this arena.
One critical enabler of Gulf Air’s new route profitability capabilities is business intelligence (BI). For a network carrier like Gulf Air, one route feeds several others, and the profitability of one route depends on others.
Everything relies on data clarity, the speed at which reports are produced and the overall accuracy of these reports in enabling airlines to make route profitability decisions.
Thanks to BI, we’re able to rapidly deliver detailed reports showing airline network flow, and the impact that one route has on the entire network. Our reports depict the structure of the network and its flow; they show dependencies between routes, and the impact that a change in frequency, capacity or timing will have on the entire network.
We achieved this for historical periods by integrating different data sources, including revenue accounting data and airline schedule data.
We wanted an open source big data solution with ease of
Dr. Jassim Haji, Director of IT, Gulf Air
installation and configuration, as well as integration with
other tools and solution maturity. We evaluated and tested
various solutions, linking them to the internet and social media,
then retrieving posts based on specific keywords in both Arabic
BI’s success depends on tools and techniques to quickly transform raw data into meaningful information. We’ve put into place a big data solution that draws on enormous amounts of data created in social media, especially Twitter, extracting what’s relevant and text mining to find the data we need.
Determining the accuracy and quality of the results proved difficult at first and it relied heavily on human intervention and perception. But during manual reviews of sampled data, we identified opportunities and implemented improvements in the logic.
One issue we needed to address was ‘data noise’. To manage it, you need strategic design and technologically sound logic, so you must produce a technological structure that removes any irrelevant data that might skew overall data accuracy and quality.
We wanted an open source big data solution with ease of installation and configuration, as well as integration with other tools and solution maturity. We evaluated and tested various solutions, linking them to the internet and social media, then retrieving posts based on specific keywords in both Arabic and English.
After our evaluation and testing, we chose Cloudera, the open-source platform for big data. Its stability, wide usage, solution maturity, community support base, ease of integration, installation and configuration made it the perfect choice for Gulf Air.
In addition, to handle the vast amount of data, we implemented the Massive Parallel Processing (MPP) concept of big data so that we could simultaneously process data on different nodes of our infrastructure.
In parallel, we defined Arabic sentiment analysis logic to include different regional dialects, formal and informal Arabic words, slang words and different spelling variations and mistakes.
Our developments to cater for languages and dialects has positioned us uniquely in the market. In fact, hardly any service providers in our region can deliver Arabic sentiment analysis, especially covering regional dialects.
Sentiment analysis of social media enables us to monitor posts related to Gulf Air, 80% of which are in Arabic. We’re only too aware of the importance of social media monitoring to understand public perception and customer opinion. With our BI and big data solution we now have an idea of our performance and this helps us to make quick decisions and find remedies.
Test and learn
We faced big challenges due to the complexity of Arabic sentiment analysis, coupled with the need to process a human language to obtain sentiment and feelings; especially with a complex language like Arabic.
To achieve a high-degree of accuracy, we applied a ‘test and learn’ approach, whereby we tested the results of the logic and reviewed their inaccuracies. Thanks to the learnings from this we could enhance the logic, build language exception rules and then test this new logic.
While it’s not possible to reach 100% accuracy, we achieved an 84% accuracy level, which according to literature is a good achievement.
We’re now able to provide end-users with seamless services regardless of whether they’re in public or private clouds.
Dr. Jassim Haji, Director of IT, Gulf Air
On top of that, we’ve expanded our cloud reach over mobile-
based clouds through mobile and tablet devices.
Cloud computing is another critical enabler for Gulf Air. The impact of cloud on an airline’s business and business continuity cannot be underestimated.
Organizations are looking at cloud as a new and revolutionary way to deploy IT services in a more cost effective manner, while maintaining high availability and flexibility in the main data center and extending this to the business continuity center.
We started by adopting a private cloud computing service to cater for internal use and data confidentiality. That meant evolving our new generation data centers, making us one of the first implementers of this technology in our region.
Based primarily on virtualization technology, storage and network components – which provide huge scalability for services – these implementations enabled Gulf Air to deploy services more rapidly in contrast to a non-cloud based environment.
From our experiences, we soon realized we needed to expand our cloud solution to achieve a wider reach, so we embraced a hybrid cloud approach. To achieve this we connected our private cloud to external services on public trusted clouds.
We’re now able to provide end users with seamless services regardless of whether they’re in public or private clouds. On top of that, we’ve expanded our cloud reach over mobile-based clouds through mobile and tablet devices.
I’d say the main driver for a hybrid cloud approach is that it’s not feasible to host certain systems in a private cloud. That’s particularly the case with complex and heavy transactional systems that need to be geographically distributed, or other systems that rely on industry collaboration.
While we’re making headway in the cloud computing area, our partnership with SITA helps us to continually drive this forward. (See the box below: Pushing boundaries)
Gulf Air has gone one step further in protecting business data by providing real-time data replication between the sites, allowing
Dr. Jassim Haji, Director of IT, Gulf Air
a total data storage size of 10TB.
Virtualization is playing a role in other areas of the business as well. With increasing demand for more Gulf Air destinations, along with our network growth, we evaluated the virtualization of our desktops. We wanted an approach that allowed us to quickly respond to business demands.
From an IT perspective, Gulf Air’s virtual desktop infrastructure has hugely impacted our business, empowering us to more easily open new offices, stations and sectors – so we can react to the business demands quickly and accurately.
Added to that, and aligned with Gulf Air’s work on business continuity, it’s imperative today to build a reliable, flexible and cost effective Disaster Recovery and Business Continuity Center. This can be achieved only as a result of successfully implementing cloud computing.
Cloud has helped in the cost-effective establishment of our Disaster and Business Continuity Center, without incurring additional running costs or increasing the number of human resources.
Traditionally, building those sites is extremely expensive and requires double the amount of resources. But with cloud computing, we can activate IT services immediately at the Disaster Recovery site by seamlessly moving it virtually from the main site, with minimal impact to the business as a whole.
Data protection is another challenge in the build of a Business Continuity Center, as most organizations require backup tapes to restore data when the site is activated. Gulf Air has gone one step further in protecting business data by providing real-time data replication between the sites, allowing a total data storage size of 10TB.
As a result, the site can be activated immediately once a decision is taken to start operations from the Business Continuity Center. In an industry where time is money, this development is one that makes Gulf Air stand out from the rest. (See the box below: Business continuity in real-time)
In fact saving time and money, while boosting performance and service, are principles that apply to many of the technological developments we’re putting into place to help drive Gulf Air’s business.