Kuoni Travel Group India's disaster recovery plan for business success

The story of how Kuoni India implemented a world-class disaster recovery setup for its global network.

India's travel industry witnessed exponential growth in the post-liberalization era, and Kuoni Travel Group India has been a leading player that capitalized on this. However, this rapid rise brought increased availability challenges for Kuoni India's CIO, Dhiren Savla, and his team.

VFS Global, a business process outsourcing firm of the group, plays a critical role in handling the visa-related data processing work of 27 diplomatic missions. It is present acrossoperates in 44 countries and processes in excess of 7 million visa applications (contracted) annually. "Need for disaster recovery came primarily on the VFS Global side in the last two years, since it's a completely IT-enabled business with very high dependency on IT," Savla says.

Due to Kuoni India's varied business models, its first challenge was the business risk associated with operating out of a single Mumbai-based data center. The group's geographically distributed businesses also translated to latency issues.

A multi-pronged DR plan

The call was taken to create a disaster recovery site to take care of the primary geographical challenge. This data center would be closer to the distant locations and reduce latency issues, whereas India would host the other locations. If a data center went down, the other would assume control to ensure uninterrupted business. A certain amount of latency and performance degradation was acceptable in such a case.

Savla and his team kicked off Kuoni India's disaster recovery project in February 2008, with workshops to capture the main business requirements. It attempted to reach an agreement on objectives like latency, high availability and better business continuity.

"It took around six months to agree on aspects such as design and location, so we finished the exercise in July 2008. We decided to implement disaster recovery in two phases, with our primary data centers in Mumbai and London. We chose London, since it's a neutral location in Europe," Savla says.

Started in September 2008, the disaster recovery initiative's initial step was to consolidate Kuoni India's IT infrastructure. This included reducing the number of servers and rebuilding the group's Mumbai data center hosted at Sify Technologies Ltd.'s facility in Vashi, Navi Mumbai. The Mumbai data center had more than 40 servers before consolidation, and 18 servers after. Additional redundancy to boost Kuoni Travel's Mumbai data center was achieved by deploying nearly identical server banks at the site, which takes care of hardware failures.

The next step was to commission the London data center, which included infrastructure and storage area network investments. Kuoni India's entire application portfolio was broken up, depending on its area of application. At present, Kuoni India has around 200 applications (140 in India and 60 in the U.K.). Savla claims substantial savings on total cost of ownership, management, hosting, software licenses and manpower due to the exercise.

In addition to the existing Multiprotocol Label Switching virtual private network (VPN) from Sify for Kuoni Travel's Indian network, a parallel MPLS VPN from Reliance is in place for agile operations. At the moment, Kuoni India is in the process of network consolidation for its international locations.

"We are consolidating the multiple networks that are in use with an N+1 type of MPLS network. It will be a VPN with a backup Internet link, because having two parallel MPLS networks globally is not feasible," Savla says. He expects to complete this section of his international network by year's end, he says.

Data migration challenges

Once the infrastructure was in place, Kuoni India's team had to migrate past data. Replication was not a feasible option at this point due to the huge data volumes. So the option was to physically move data on storage devices.

Due to the distance between Mumbai and London data centers, it would take 48 hours for the physical relocation of data. In the meanwhile, high volumes of data being generated by the business had to be replicated. Savla and his team decided to physically move data, and then replicate data generated during that time.

Go live

Kuoni India's disaster recovery exercise was complete by February. Transformation, migration, consolidation, disaster recovery and latency issues were ironed out during this period within budget.

After a full-fledged disaster recovery test, the entire setup was operational in March. Today, the Mumbai data center serves more than 400 Kuoni India locations, more than 300 VFS Global offices (around 65% of VFS Global locations worldwide) and the entire group's email services. Mumbai data is replicated to London and vice versa, with around a minute's delay. According to Savla, the automated switchover between data centers has been thoroughly tested (with a switchover time of not more than five minutes in case the automatic switchover fails). The data center has a recovery time objective (RTO) of around 10 minutes.

Quest for higher RPO and RTO

Savla will thoroughly test the existing disaster recovery setup's stability and maturity with its first disaster recovery (DR) drill in July 2009. He says he also plans to improve the DR setup to achieve higher recovery point objective and RTO levels with two near DR sites. The first near DR site will support the Mumbai data center, and is likely to be in Chennai. The second near DR site will support the London data center and possibly be in Belgium.

Read more on Disaster recovery