Storage Special: Case Study

The airline that’s never grounded: IA’s DR saga

59 aircraft, 66 offices countrywide, 17 offices overseas, approximately 275 flights flying 20,000 people a day and 210 GB of mass storage. These are the components that go into the making of domestic air carrier Indian Airlines. IA is probably one of the few public sector organisations with a disaster recovery system up-and-running. Rahul Neel Mani reports

Cost, time and business process continuity are the major considerations in any DR implementation, says A k Rastogi

At Indian Airlines (IA), information technology is not just supporting business, it’s spearheading it. The continuity provided by IT is crucial to run the IA business effectively. For A K Rastogi, director IT at Indian Airlines, planning for business continuity is a priority focus area at present.

Considering the potentially devastating impact of any disaster, Rastogi’s aim is to ensure IT infrastructure at IA is robust enough to withstand any calamity. "IT infrastructure has become harder to manage and control as it is distributed across the enterprise," says Rastogi. IA’s disaster recovery plan involves the integration of various hardware, software and network equipment. "Various components are configured as per business process continuity requirements, which depends on the criticality of the business function," elaborates Rastogi. He feels that the cost of a DR design would increase exponentially if one were to lower the recovery time. Therefore, cost, time and business process continuity are the major considerations in any DR implementation.

Some business process operations cannot afford any interruption, and it may be mandatory to have a hot standby configuration of software, hardware and network equipment, to ensure uninterrupted flow in the operation. Here, the recovery process is transparent to the end-users. IA’s DR plan model is based on this simple but smooth mode of operation.

Says Rastogi, "In case the business can accept shorter duration of interruptions, alternate or fallback systems may be configured to take care of DR. The database and library are maintained on primary or secondary storage an the backup system, depending on the time available for recovery and cost considerations." The priority and the costing of the solution selected influences the decision making process.

IA’s IT set-up

As Rastogi puts it, Indian Airlines is perhaps the first commercial organisation in India to have implemented a DR set-up on a mainframe system, for mission-critical applications. There are two IBM ES 9000, 9672 R21 mainframe systems with dual CPUs, placed at two different locations—about 500 metres apart in the same campus. These are called System A and System B respectively. Both the locations/systems are connected to all IA offices in India and abroad, through dedicated communication lines. The two sites are interconnected through high-speed ESCON (IBM enterprise system channel to channel connection) fibre optic links. "The system processes around 2.16 million passenger-related transactions daily. The database also has the synchronous mirror image at both the sites, ensuring 100 percent database recovery in case of any unforeseen incident," says Rastogi. Also, for other on-line business transactions, important files are captured at both the sites. Both system and network configurations are designed to ensure complete recovery of core business applications in the event of a disaster. In case of a disaster, production system applications can be run on the backup system at Site B. Data synchronisation is transparently supported by the operating system and transaction processing middleware.

At present, System A is being used fully for passenger services applications, while System B is used for various batch/online applications. It also acts as a back up to System A, with all program libraries loaded in it. "Any update to the program source and libraries are also simultaneously carried out on both the systems," says Rastogi.

Both the systems are being run round-the-clock in a controlled environment with a redundant backup of UPSs and diesel generator sets. The systems provide batch, online and real-time processing environments. Rastogi says that the resources on both these systems have been configured for optimum utilisation, keeping in view the need for backup of mission-critical applications.

"To optimise costs, IA has taken a single license copy of the mission-critical applications and associated system software with single operation at any of the sites," explains Rastogi.

The network

Indian Airlines, like any other IT savvy company, has a top-of-the line nationwide computer network. Both the Systems A and B help the airline staff access various applications using this network. The network is primarily connected by leased lines. "A mix of dumb terminals and PC workstations are used to access applications. The system has access worldwide (IC international stations and CRS systems) through SITA network on a TCP/IP-based network," says Rastogi.

There are two IBM 3745 front-end communication processors connected to the hosts to form a redundant network configuration. The two 3745 communication controllers across the sites are connected through a fibre optic link. All the devices in the network access the host via the 3745 processor. "A router-based mash network has been implemented, using a mix of 64 Kbps digital links and 9.6 Kbps analogue links, with a backup on ISDN/PSTN," he says.

Application-wise mainframe storage needs of IA

Applications and data at IA are stored on 9394 (RAMAC II) storage devices. "9394 is a rack-mounted disk array subsystem. The basic components are storage rack, the 9395 (DASD) and cluster cards," Rastogi says. The storage device supports RAID 5 data protection system. There are 4 DASD/drawer, each drawer having formatted capacity of 11.35 GB. There are 10 drawers (40 DASDs) in System A and 12 drawers (48 DASDs) in System B.

DR solution at IA

The resources available on the two systems have been configured to process mission-critical applications. The passenger services-based applications are processed on System A and the other non-critical applications are processed on System B. In the event of any disaster, System B resources are reconfigured to run mission-critical applications and process non-critical applications in degraded mode at off-peak times. The total blackout period on account of the DR implementation is around 40-50 minutes and there is no loss of data. "The DR system has been activated twice since the system was implemented, due to unavoidable contingency caused by weather," says Rastogi.

IA’s IT backend