Tracking 6,000 Trains Traveling Between 9,000 Stations
Indian Railways is the largest rail network in Asia and the world’s second largest under one management entity, spans 9,000 stations and carries 17 million passengers on 6,000 trains every day. The National Transportation Enquiry System (NTES) deployed by Indian Railways is an integrated system that allows passengers to access information on the movement of the 6,000 trains. NTES can be accessed by the public via the Web as well as at terminals at major stations. The public can also call into four call-centers staffed by customer service representatives that relay train arrival and departure information over the phone.
When designing the technology components of the system, Indian Railways realized a key requirement would be the ability to replicate data throughout the system to a mix of databases in five geographical zones across the country. "With the amount of data and transactions required to keep track of 6,000 trains, we knew we could not run train status queries against one central database," says Suneeti Goel, Senior Additional General Manager for CRIS. "We would not have been able to provide fast responses – whether the public was checking the status of a train on their own via the Internet or whether one of our customer service representatives was checking for them over the phone."
Indian Railways designed a solution whereby data would be replicated to servers accessed by the four call centers as well as a public Web site that passengers can view from the Internet or via terminals at major train stations. Call centers customer service representatives rely on an Oracle database to access the information while the Web site runs queries against a Microsoft SQL Server database.
Back-End Database Required to Support Heterogeneous Front-End Databases
"To support these databases, we needed a robust back-end system with a Sybase ASE database in five zones from where the information is replicated to 4 other Oracle servers (call center) and from these four ASE databases, information is also replicated (custom code written in .NET) to Microsoft SQL Server (for the website) – one for each of the four call centers and a fifth one for the Website. Suneeti says, "The database needed to integrate with Oracle and Microsoft technology, and it needed to possess the ability to constantly replicate data across the system in real-time so that train updates would appear immediately on all of the databases."
To take on this challenge, Indian Railways turned to Sybase for three key technology components: Sybase ASE, Sybase Replication Server and Enterprise Connect Data Access. "Sybase technology was ideal for our situation because of its flexibility to integrate with databases from Oracle and Microsoft as well as the reliability of its replication," Suneeti says.
Data on train movement is generated by a control charting application running in 72 locations throughout India. A TIBCO solution integrates the charting application with Sybase Replication Server to upload the information into the Sybase ASE server network within each of the five zones of the Indian Railways system. Each instance of ASE runs on a Linux operating system and a combination of IBM System X servers and AMD servers. Indian Railways also deploys an ASE warm stand-by at all five zones providing high availability in case a server needs to be taken down for maintenance or an unplanned emergency.
Data is replicated bi-directionally by Sybase Replication Server in real-time among the five ASE servers so that each server has a local, up-to-date database showing the status of all 6,000 trains. Replication Server constantly replicates 10 tables bi-directionally among the ASE servers, executing on average 14 transactions per second that are usually 200 bytes large. "Our transactions per day sometimes reach as high at 1.2 million," says Sukh Dayal technical team member of NTES, CRIS, "And we sometimes reach as high as 60 transactions per second. Sybase Replication Server handles this rate flawlessly."
Indian Railways currently transfers data from the ASE network to Microsoft SQL Server using .NET custom code but will eventually change to replicating with Sybase Replication Server. To replicate data to the Oracle databases at the call centers, Indian Railways relies on Sybase's EnterpriseConnect Data Access (ECDA) technology. As an add-on to ASE, ECDA enables ASE to provide a consolidated view of enterprise data across multiple heterogeneous data servers. With ECDA, the Indian Railways Oracle database constantly accesses data from the ASE servers as though all the data was stored in at single database.
Simple to Deploy and Offers Intelligent Capabilities
Indian Railways deployed the Sybase technology using its own IT staff with assistance from Sybase Professional Services, which assisted with the loading of the Sybase operating systems. "We were able to handle most of the deployment on our own, but Sybase helped by providing clear documentation on the Internet for troubleshooting, development and implementation," says Suneeti.
Sybase helped Indian Railways configure the solution that has performed 24x7x365 for the past eight years and lends itself to future enhancements. Because Sybase applications are light, they also require fewer system resources such as RAM and CPU, which helps make NTES response time faster so customers can immediately find out the status of trains.
"Replication Server is very intelligent in that it lets us use table-level replication for transactional data so we can optimize bandwidth, and it guarantees the transfer of data with low latency," says Sukh Dayal "The speed helps us maintain a cost-effective network, and Replication Server integrates well with data from homogeneous systems as well as heterogeneous environments, which is unique feature."
Suneeti adds, "Sybase technology requires very little maintenance. We soon plan to upgrade the system to ASE 15, a process we expect to last just a few hours. With ASE 15, we can further optimize the system to improve the performance speed since ASE 15 handles large databases so well. We also want to take advantage of its partitioning, encryption and clustering features."