Publish Date:2010-01-01Summary:Karnataka's legendary bank, with an 85-year-old history and a sprawling database of 4.5 million customers was ailing with a weak disaster recovery plan. Read how the bank went in for a massive IT revamp to ensure full data security to its customers.Reader ROI:The impact of disaster recovery on a business makingHow a three-tiered storage solution can helpThe importance of scalabilityTHE ORGANIZATION Founded on the coastal banks of Mangalore, Karnataka Bank has become a name to reckon with in India\u2019s financial sector. With an 85-year banking history and a network of 456 branches, the bank has become synonymous with reliability to many of its 4.5 million customers. But with that trust comes responsibility.\u00a0 \u201cOur banking environment requires real time access to information with no room for data loss,\u201d says Ananthakrishna, non-executive chairman, Karnataka Bank.\nWatch CIOs in conversation with business leaders\u201cData lost in transit is irretrievable. This would seriously compromise our credibility.\u201dThe Business Case Handling and ensuring the safety of a sprawling database of 4.5 million customers is hard. So much so, that the Reserve Bank of India passed a regulation two years ago, urging banks to conduct frequent disaster recovery drills to ensure zero loss of customer data. Karnataka Bank had a disaster recovery process it place, but it wasn\u2019t the best.\nAlso readThe Top Technology Priorities\n\u201cIt would take four to five hours just to initiate the recovery\u00a0 process at the DR site,\u201d says Anantha Padmanabha Upadhya, assistant GM IT at the Bank. \u201cWhen we conducted DR drills, the results revealed glaring loopholes.\u201dA recovery time objective of four hours was become unacceptable to the bank and Upadhya knew something had to be done. It wouldn\u2019t be an easy problem to solve. The bank has multiple delivery channels including its branches, ATM, website and SMS banking and each channel housed its data separately. As the bank grew, the silos of data made consistency hard to achieve. Data back up and restoration operations also took hours and brought the bank\u2019s database to a halt, affecting core banking operations. Upadhya had another worry. A poor recovery point objective left a window open for data loss. \u201cData lost in transit is irretrievable. This would seriously compromise our credibility,\u201d he says. Upadhya realized that the problem lay at a fundamental level. The existing DR setup involved data replication at the application level, resulting in multi-application data copies. The bank\u2019s applications, which included core banking, were\u00a0 slowing down the replication process. \u201cWhat we needed was a consolidated IT infrastructure at the primary datacenter to facilitate centralized data replication at the DR,\u201d explains Upadhya. This would reduce complexity, time and effort and shut out the possibilities of\u00a0 human error.\u201cData lost in transit is irretrievable. This would seriously compromise our credibility.\u201dThe Project- Weighing all his options, Upadhya turned to a three-tiered storage solution. The new setup created a bunker site between the primary datacenter and the DR site. This near-line datacenter operates synchronously with the main datacenter so that every transaction is updated in real time. The remote datacenter, in a different city, operates in an asynchronous mode, completing the set up. This three-way DR solution creates a foolproof strategy against any sort of disaster. The near-line datacenter is located in the same city as the primary datacenter to avoid any latency, since data is transmitted synchronously to both sites. When the primary datacenter goes down, the near line DR datacenter takes over and starts syncing with the remote datacenter and incrementally replicates every last change to the main DR site. The system uses a universal replicator that can work in a heterogeneous environment and with multiple applications. In addition, the re-sync feature provides the bank\u00a0with concurrent multi-site replication capability and enhances data protection. Further, a split-mirror mechanism provides a rapid disk mirror of production data by copying only incremental changes within the local storage system. The process rapidly creates multiple copies of critical data that can be recovered in a disaster. In addition, it enables the logical partitioning of ports, cache and disk capacity. The system has built in redundancies at every point to ensure a resilient storage based DR infrastructure.\nThe First Steps While the plan looked good on paper, the tough part was executing to deadline. The bank was up for year-end processing in March, which left Upadhya with three months to implement and stabilize the project. The challenge was amplified because the project had to be implemented on a core banking system and the migration had to be seamless without disruptions. Faced with these challenges, Upadhya called for back-up in the form of an outsourced partner. In addition, he would upgrade his servers within a few months to optimize the benefits of the new storage infrastructure. \u201cWe invested around Rs 10 crore in the project. I believe we are the only bank in India which has a DR plan of this scale and nature,\u201d he says.The Bottom Line Today, Karnataka Bank can assure it\u2019s customers of zero data loss. With faster backups and data replication, the bank has reduced its recovery point objective to zero, and reduced its recovery time objective from about five hours to 15 minutes. Data backup and restoration processes have been cut from four hours to a few minutes. The efficiency this has triggered has seen the bank\u2019s performance improve by 20 percent in the last year. And the system can support the bank\u2019s growth over the next five years, without any fresh investments.