Airline's disaster recovery strategy boosts data centre efficiency

Flybe uses data backup and replication software in a disaster recovery project that improved data centre efficiency and won Best Virtualisation for Disaster Recovery Project award.

This article can also be found in the Premium Editorial Download: IT in Europe: The realities of virtualised data centres

Exeter-based airline Flybe had native backup and replication software that couldn’t support a growing virtualised infrastructure or the company’s key business operations. Flybe needed a scalable backup tool for its virtual machines (VMs) that would run critical applications, such as aircraft maintenance apps.

“We needed a flexible solution with low support overhead,” said James Richards, Flybe’s virtualisation and server specialist. “It had to have recovery options for applications like [Microsoft] Active Directory and Exchange, and we wanted the ability to automatically test our restores.”

Flybe's IT infrastructure includes 268 physical and virtual servers and desktops. Its VMs run its website (which processes flight searches and reservations) and business applications. “Each of these virtual machines is critical for day-to-day operations that drive revenue, so downtime is not an option,” Richards said. That’s why the company launched a disaster recovery strategy.

Issues of native backup and replication software
Its existing backup and replication software was limited to  creating just 500 GB backup data stores. It also did not offer application-level recovery or automatic backup verification.

“We had to manually check backup jobs every day,” said Richards without revealing the name of the native software. “It is fine if you are running a small number of VMs, but it just wasn’t conducive to a large and growing virtualised environment like ours.”

In March 2010, the IT team embarked on a virtualisation and disaster recovery strategy. The team identified several infrastructure concerns:

  • The environment needed a server refresh because of aging end-of-life hardware. Commissioning a new server took several days.
  • Services were tied to underlying physical server hardware, making disaster recovery difficult.
  • The airline was running out of data centre space.

The team’s goal was to reduce power and space requirements of the server estate. “Virtualisation offered the consolidation of ageing servers on new energy-efficient servers,” Richards said.

Prior to the disaster recovery project, Flybe’s infrastructure hosting its e-commerce website and internal company services was entirely physical. “Our backup method for this environment consisted of backup agents running on each server OS,” Richards explained, adding that this process was time-consuming and less efficient.

Data backup and disaster recovery strategy

The team recognised that backup options for a virtualised environment are quite different from those for a physical one. The old tools and techniques -- designed for an exclusively physical universe -- would not scale to a substantial number of servers.

After reviewing several backup and replication technologies for its VMs running on a vSphere platform, the team selected Veeam Backup and Replication 5 [which won best new technology at’s Best of VMworld Awards 2011].

“Veeam offered features using VMware technologies, specifically VMware change block tracking, Veeam SureBackup and Veeam Instant VM Recovery,” Richards explained.

Flybe’s IT team implemented Veeam’s technology to back up production VMs daily and to test VMs weekly. “We couldn’t perform daily backups on this many VMs in the past because the backup window was just too long,” Richards said.

The implementation involved the install and configuration of a new storage, network and server layer.

He demonstrated how the team uses every feature of the tool. “Every morning I receive an email report from a Veeam SureBackup job verifying the backup of our Exchange 2010 environment,” he said. “When there [is] a problem with one of the VMs that runs an important aircraft maintenance application, I kick off Instant VM Recovery and users [can] … access the application within minutes,” Richards said.

Flybe also uses Universal Application-Item Recovery (U-AIR) to recover individual objects from any virtualised application. “We use it for Exchange email restores,” Richards said. Previously, the team had to stage the entire database to disk and then retrieve individual items and restore them to users’ mailbox.

Benefits of new backup and replication software

The product offers restore options at all levels, including for VMs, guest OS files and individual items running in applications such as Exchange and SQL Server, he added.

Critically important, the disaster recovery project has improved the team’s ability to react to business change. “We are able to deploy, configure, protect and recover applications on a virtualised platform much quicker than when they were on a physical platform,” Richards said.

A full VM restore is far faster now because of vPower, a Veeam feature that allows a VM to run directly from a backup file, he explained.

In addition to the backup and replication product, Flybe’s current IT infrastructure includes VMware vSphere Enterprise Plus 4.1; VMware Site Recovery Manager 4.0; Hewlett-Packard ProLiant DL580 servers; HP ProCurve 5412ZL switches; HP EVA 8400 storage area networks; and HP Continuous Access and Business Copy EVA software .

As new virtual machines are created, they are automatically added into the correct backup jobs – a major benefit for a growing architecture. “This is done by targeting VMware folder structures from Veeam,” Richards said.

The IT team’s strategic approach in addressing the challenges of virtualised infrastructure and in replacing its native backup and replication software impressed the judges of TechTarget’s Best of VMworld Europe 2011 user awards. Flybe won the Best Virtualisation for Disaster Recoveryproject award at the VMworld Europe 2011 conference in Copenhagen, Denmark.

“Once Flybe implemented [Veeam’s] solution,” a judge noted, “each of the features made the life of the VMware administrator, Exchange administrator, end user, and even customers much improved.”

Now Richards wants to finish virtualising Flybe’s remaining 40 physical servers hosting internal company services and use this project as an opportunity to upgrade to Windows Server 2008 R2.

Click here to find out about other winners of the Best of VMworld Europe 2011 user awards.

Check out all our VMworld Europe 2011 conference coverage here. 

Archana Venkatraman is the site editor for Write to her at [email protected].

Read more on Datacentre disaster recovery and security