Case Study: United Airlines Boosts IT Efficiency, Business Resiliency with Private Cloud Solution
To reduce operating costs and find more effective ways to attract and retain customers in the hyper-competitive airline industry, United Airlines is working with Microsoft to expand private cloud computing to the enterprise. United Airlines uses Windows Server 2012 R2 Hyper-V Replica and Microsoft Azure Site Recovery to expedite the migration and consolidation of virtual machines and mission critical services to its new data center in Chicago and provide high availability services when hardware fails.
“To build an eight-node cluster in cooperation with the storage and networking teams used to take three to five days. Now the operations team can do it all themselves in a half-day.” Richard Wilson, Principle Architect United Airlines
Following its 2010 merger with Continental Airlines, United Airlines is reaching new levels of customer satisfaction. Headquartered in Chicago, Illinois, the company was rated the world’s most admired airline on Fortune magazine’s 2012 airline-industry list of the World’s Most Admired Companies.
Behind the accolades, the United IT team works to keep the underlying reservation system, baggage-handling system, public website, aircraft maintenance records, database servers, and countless other computer systems running flawlessly around the clock. It is hard work and it means that the IT team is constantly looking for ways to deliver new services faster and to streamline both its capital outlay on servers and the operational expenses involved in running them.
“The airline industry in incredibly competitive,” says Eric Craig, Managing Director of Enterprise Architecture at United Airlines. “We have the most comprehensive network on earth, but that’s not enough. We have to earn our customers’ business on each and every flight with the great on-time performance, excellent customer service, and innovative features our customers want. Our IT infrastructure needs to be reliable, cost effective, and highly adaptive so we can invest more capital into our customer facing products rather than data center servers.”
Early Virtualization Projects
In 2008, United made significant progress in building a cost-effective, flexible, and scalable IT infrastructure by virtualizing its data center infrastructure. Virtualization not only helped United to reduce IT costs but also to improve business agility—the ability to respond to business needs faster by deploying virtual machines in hours rather than weeks.
United even virtualized its business-critical United.com website and reservation system, which was running on physical servers. “United.com is an incredibly important channel for us,” Craig says. “Between 30 and 40 percent of the airline’s revenue comes from United.com—in excess of [US]$10 billion annually—so it needs to be running on a resilient, adaptive, and scalable infrastructure.”
At around the same time prior to the merger, Continental had virtualized about half of its Houston infrastructure, using the Hyper-V technology in the Windows Server Datacenter operating system.
Merger Provides Incentive for Cloud Computing
Following the 2010 merger with Continental Airlines, United Airlines determined that the airline should continue along the virtualization path and adopt private cloud computing, which encompasses reliable, scalable, on-demand compute, storage, and networking services and offers point-and-click resource provisioning for business units and self-service provisioning for the software development organization.
“With the automated management capabilities of private cloud computing, we are responding to business needs to increase compute capabilities, or to improve performance to minutes instead of days,” Craig says. “The ability to move applications from one server to another without disrupting your business, or to roll back an installation when you deploy a product that didn’t quite work out—that kind of adaptability and manageability is a great cost reducer and business enabler. Private cloud computing is the only way to get from 50 servers per administrator to 1,000 servers per administrator.”
United.com IT staff wanted to ensure its infrastructure had the resilience, scalability, and manageability required to boost the airline’s competitive position in the aviation industry. The IT team also wanted to have high availability as a foundational feature of its new data center to minimize downtime for its business-critical systems.
“We can’t afford any service outages, so it was important to pick the right high availability and disaster recovery (DR) solution: resilient, flexible, easy-to-use, and cost-effective,” says Richard Wilson, Principle Architect, Microsoft Private Cloud and Windows Server at United Airlines. “Our existing manual failover scenarios and expensive storage arrays were from multiple vendors and they were complex to manage.”
At the same time, United wanted to consolidate its data centers. It decided to close its Houston, Texas-data center facility and it needed an efficient, automated method to migrate its virtualized Microsoft infrastructure in Houston to its new, more cost-effective data center in Chicago. Each of these scenarios would benefit from a cloud computing solution.
“To exploit the power of cloud computing, we needed a partner that offered more than just a powerful hypervisor,” says Wilson. “We also needed a comprehensive management tool that we could use to manage the cloud fabric, from the physical servers to the virtual machines, storage, and networking environments.”
United Airlines is using the Windows Server 2012 operating system, including Hyper-V virtualization technologies, as part of the technology stack for the private cloud. “Microsoft technologies are easy to work with, interoperable, flexible, and cost-effective,” says Wilson. “As such, we see the Microsoft private cloud as a strategic enabler to streamline our integration efforts, and as a way to reduce the cost and complexity of the merger.”
Microsoft System Center 2012 data center solutions serve critical roles in United’s private cloud strategy and work well with other management tools. “United Airlines is a large, technically heterogeneous and complicated enterprise. No one tool can provide everything we need,” says Craig. “System Center is an important component of our management stack, providing orchestration, provisioning, migration, and automated recovery services throughout a large portion of our IT landscape. Microsoft recognizes the management challenges of large enterprises and has ensured that System Center interoperates well with the rest of our tool stack.”
Working with Microsoft
Part of the decision to choose Microsoft technologies lies in the close working relationship that has developed between United and Microsoft. Before the merger, both Continental and United had participated in many Microsoft Rapid Deployment and Technology Adoption Programs. Post-merger, United joined the Technology Adoption Program for System Center 2012.
“This was a great opportunity to work with the product team and gain early access to the features, the ability to have input and shape what the product will look like, and to get features that we really needed,” says Wilson. “Over the last couple of years, we have been extremely happy with the support and knowledge of Microsoft Services Consulting, which has yielded some IT highlights post-merger.”
Migrating United.com to a Hyper-V Private Cloud Environment
One of these highlights is a joint Microsoft and United project to accomplish the migration of United.com to a Hyper-V private cloud environment. In March 2012, with the private cloud up and running, the United IT team pulled off what Craig calls “one of the most complicated, massive cutovers in transportation history”— which included moving the business-critical United.com website from a physical server environment to a Hyper-V cloud environment.
“We did continuous system testing prior to the cutover date—both scale and functionality testing,” Craig says. “We also worked with our business partners to ascertain all the scenarios that were likely to drive traffic patterns up or down during the migration.”
As the Microsoft and United team members were modeling those scenarios, they discovered that they didn’t have enough servers supporting United.com to respond to worst-case scenarios. They quickly solved this issue by taking advantage of automated build development.
“Just two or three days before this incredibly important event, we deployed enough Hyper-V virtual machines to support the site, and we did it in hours using automated server builds and application deployment,” says Craig. “Physical host builds used to take days to complete and a 30-page manual document. We reduced this to 2.5 hours with automation. We would not have been able to respond in such a short time without the private cloud technologies we had in Windows Server 2012 and System Center 2012.”
Building a Disaster Recovery Solution
To address the need for an enterprise-ready disaster recovery solution, in June 2013 United Airlines joined the Rapid Deployment Program (RDP) for Windows Server 2012 R2. “Now that we are more virtualized, we are looking at a whole new approach to DR, where flexibility and cloud computing combine to provide a resilient solution that we can tailor to meet our needs,” says Wilson. “It made sense to continue on our cloud journey with a Microsoft DR solution.”
Hyper-V Replica offers a data replication solution that replicates virtual machines within a site or to a remote site. The latest version of Hyper-V Replica provides the flexibility that United is looking for, with variable replication frequency—from 30 seconds up to 15 minutes—and support for extended replication to a third site. And the new DR management service, Microsoft Azure Site Recovery, answers the airline’s need for a highly available DR solution because it is delivered as a cloud service running in the Microsoft Azure environment. Azure Site Recovery offers orchestration at scale delivered via recovery plans, so United IT staff can bring up applications in a desired manner at a low recovery time objective. While Azure Site Recovery is a feature of Windows Server 2012 R2, it supports backwards compatibility with all versions of Hyper-V Replica.
Enabling Data Center Migrations
During the RDP, the IT team realized it could use Hyper-V Replica and Azure Site Recovery for the migration of virtual machines from Houston to Chicago. “We were excited by this unusual use-case scenario, which underlies the flexibility of Microsoft technologies,” says Wilson. “Being able to take advantage of these technologies to migrate services sets a DR solution from Microsoft apart from other solutions available in the market.”
In Houston, the IT team deployed servers running Windows Server 2012 R2 as a pre-migration environment and installed a 1-gigabit circuit between Houston and Chicago for the replications. “We’ll use Azure Site Recovery to initiate the failover so that the virtual machines will become live in Chicago,” says Wilson. “When we complete the Houston migration, we’ll use the same solution to replicate non-production systems from our data center in Charlotte, North Carolina, to our center in Chicago. Then, we plan on using Hyper-V Replica and Azure Site Recovery as a cost-effective DR infrastructure between our data centers.”
United Airlines is using Microsoft virtualization technologies to streamline its integration efforts, reducing the resources required to consolidate its IT environment following the merger. At the same time, the company is creating an agile, responsive, cloud-based IT environment that will help build long-term adaptability and resilience in the highly competitive aviation industry.
Improves Business Agility, Customer Service
Even when it comes to the vagaries of the global airline system, United is using private cloud computing to accommodate fluctuations in site traffic on United.com and keep its customers happy. “The airline industry must respond to unpredictable global events in real time—natural disasters, security threats, changes in travel demand caused by other transportation sectors—all sorts of things will happen all over the earth that can send customers to our site, and we won’t be able to predict the traffic patterns,” says Craig. “We also have to accommodate predictable events, such as marketing campaigns or fare sales. The point is, we need to scale United.com quickly and dynamically. With System Center 2012, we can automatically match computing power to website traffic. That’s IT that truly supports the business.”
The key to winning sales and driving consumer loyalty is the ability to offer more competitive online services for customers. United Airlines intends to use System Center 2012 R2 in its development environment to improve business agility by reducing time-to-market and introducing competitive online services before other airlines. “What’s the duty of an infrastructure team to the rest of the business? It’s to provide a resilient, scalable, rich, flexible infrastructure so that when the business comes to you and says, ‘I want to implement a brand new application,’ you can roll out those new technologies quickly, easily, and cost-effectively,” Craig says.
Reduces IT Costs
United Airlines stands to save millions of dollars in data center costs through private cloud computing. Aiming for 100 percent reliability of United.com, the company used to buy more computing capacity than it needed to have capacity in reserve. However, availability through redundancy was expensive. With cloud computing, the cloud fabric flexes to absorb traffic bursts, and workloads move around the cloud dynamically to make maximum use of resources.
Cloud computing also lowers the cost of rolling out new services. “We had all these physical servers, and before we deployed anything new, we would take some of them out of service, deploy the new application on them, wait a couple of days to see if it was OK, and then bring the service online,” Craig says. “This was inefficient from a capital allocation perspective. We needed something less expensive and more dynamic. Using private cloud computing is a far smarter approach.”
The engineering team that is using System Center to develop applications for Microsoft SharePoint is saving labor, power, and rack space costs by building out the collaboration environment in the cloud. “To build an eight-node cluster in cooperation with the storage and networking teams used to take three to five days,” says Wilson. “Now the operations team can do it all themselves in a half-day. That’s just one small group; when this development approach spreads across the company, efficiencies and cost savings will increase exponentially.”
Supports Business Continuity
Deploying the latest business continuity solution is the airline’s most recent step forward in its cloud computing journey with Microsoft. United is using its new disaster recovery solution to achieve the following benefits:
Multipurpose solution provides extra value. While peace of mind is a significant benefit, a DR solution can represent a lot of IT resources sitting in readiness on the shelf. This is not the case with United today. “We’re using our Microsoft DR solution to expedite a key operational project—migrating our Hyper-V virtualized environment from Houston to Chicago—while reducing risk and management overhead,” says Wilson. “The faster we get our workloads to run in the more efficient Chicago facility, the faster we can start to reduce our data center overhead.”
Resilient disaster recovery reduces downtime. On-premises DR software is susceptible to the disasters that can hit a data center. But no matter what happens on the ground, United IT staff can always access their Azure Site Recovery panel through an Internet connection. “With Azure Site Recovery, we have an always available management panel to enact our DR plans as soon as possible, reducing downtime,” says Wilson.
Reduced costs. United had already shipped several sophisticated storage arrays to Houston to use for the replication, but now it can repurpose that investment for other purposes. “Hyper-V Replica and Azure Site Recovery will allow us to use lower-cost storage platforms and still get the resiliency we need. This solution will save us a lot of money,” says Wilson.
Simplified recovery orchestration reduces IT management. The IT team is confident that the new high availability and DR solution won’t be a drain on their time. “From what we have seen, this isn’t going to be a system that will be difficult to set up and support,” says Wilson.
Increased flexibility saves bandwidth. “With flexible replication intervals, we can reduce replication times for critical systems, such as reservations, and save bandwidth by allotting longer replication intervals to a system that isn’t used as frequently,” says Wilson. “We are excited to put our new solution from Microsoft into production.”
Concludes Craig, “Because our industry is so incredibly cost-sensitive, it’s essential that we’re getting every penny’s worth of value out of every IT asset we have in our enterprise. We can’t waste money on spare server capacity or data center costs. With a Microsoft private cloud solution, we are able to reduce our IT costs dramatically.”
Transform the data center
The hybrid cloud from Microsoft transforms the data center by extending existing investments in skills and technology with public cloud services and a common set of management tools. With an on-premises infrastructure connected to the Microsoft Azure platform, you can deliver services faster and scale up or down quickly to meet changing needs.