Sometimes, it takes more than a single disaster recovery (DR) site to counter disaster—one of the reasons why Corporation Bank chose to add to its disaster recovery setup. Corporation Bank, founded in 1906 in Udupi, is one of the oldest Indian public sector banks. It currently operates from branches across India, with more 12,000 employees. With growing business volumes, though the bank already had a primary DR site at its headquarters in Mangalore, in 2009 it decided to go in for a nearline DR setup.
The bank's data center is located in Bengaluru. It chose to have a nearline disaster recovery setup about 40 km away from the data center. Explains S Kumar, assistant GM, IT, Corporation Bank, "If a disaster strikes at the data center, I will be in a position to quickly operate from my nearline disaster recovery setup. It is much easier to transfer data from the nearline site to the data center, than from the DR site." The data center and the nearline site are linked by high bandwidth connectivity. External consultants were involved in searching for a site for the nearline facility, and its set up.
Essential care was taken in terms of selecting sites for both the data center as well as the nearline disaster recovery setup. Says Kumar, "We had to keep three things in mind. First, that neither facility should be located in a prominent place which is susceptible to attacks. Second, that neither of the sites should be susceptible to flooding. Third, that there should be availability of power all the time, and that it should be possible to get power from two different grids." As of now, there are sufficient UPS systems to take care of the power requirements. Corporation Bank has redundant generators for uninterruptible power supply. N+1 redundancy is being used.
In all the places, enterprise-level storage subsystems have been installed. The storage in use is from Hitachi Data Systems. For all transactions that are taking place, synchronous replication is performed between the data center and the nearline disaster recovery setup. The disaster recovery setup uses a leased line as well as MPLS cloud on the connectivity front. All the replication is storage-based and not host-based; thus, box-to-box replication is taking place. The lag between the data center and disaster recovery setup is virtually zero, because a bandwidth of around 45 Mbps has been provided.
Corporation Bank's disaster recovery setup is an exact replica of the primary site. According to Kumar, "The only difference between these two is that there is site-level redundancy at the data center." The data center's two servers are in active/passive or active/active mode. The bank relies on the Sun server platform. The DR site has a single server for the core banking system application, while there are two servers at the primary site. Comments Kumar: "Right at the inception stage, we chose to design an extremely lean data center architecture, rather than build a server farm." The recovery point objective (RPO) for the disaster recovery setup has been finalized at 30 minutes, and recovery time objective (RTO) at an hour.
When it comes to security, physical access, logical access and network access have been ensured. Apart from that, Corporation Bank recently went in for an e-digital signature initiative. On the DR front, a regular DR drill is ensured every six months. "Right now, we are happy with our disaster recovery setup," says Kumar. "Future customer needs will guide us to the next move."