Internet data dissemination
  • Summary

    Industry Financial sector/ capital markets

    Client The National Securities Depository Limited, India's largest securities depository

    Requirement Clearing Members of the depository had to settle their securities transfers by a given cut-off time after trades were completed. If they did not have sufficient balance in their pool accounts, they would default during settlement, leading to resolutions of such defaults, monetary loss and penalties. Therefore, they were always under pressure to track their pool account balances to ensure that they did not default. The core depository systems were not Internet-accessible, for security reasons. A new system was needed which would publish this position data in near real-time with appropriate access controls.

  • Our solution: SPEED

    Our solution was called SPEED (Securities Positions Easy Electronic Dissemination). It was designed to eliminate any kind of direct connection from the Internet to the internal business systems, at the same time allowing frequent publishing of data from those systems to the Internet.

    Business benefits In the absence of SPEED, Clearing Members would call up either their depository participants (DP) or NSDL directly on customer-service lines and inquire about the pool position status very frequently. With the deployment of SPEED, the load on customer-service on NSDL disappeared. Clearing Members also found it easier to maintain appropriate account balances and avoid defaulting at settlement.

    Design The system had the following components:

    • A module on the core system which dumped data in text format every couple of hours, on a shared drive of a file server.
    • A module on the SPEED server which read these files, checked for correctness and consistency, and updated the positions in its internal database, discarding earlier positions data.
    • A browser-based interface with user authentication, allowing each counterparty to see his own securities positions only.

    With this stack, SPEED was able to deliver Internet-visible updates to securities positions within two hours of updates happening in the core systems.

    The system supported various types of files to convey various types of information, including pay-in and pay-out transactions, corporate actions, inter-depository transactions, etc.

    Technologies The application was developed in Perl. User interfaces were Perl/CGI programs. Sybase on Linux was used as the database. The OS was SuSE Linux v6; the system ran on x86 servers.

    Deployment and extensions The initial design of the system involved manual transfer of files from the core depository system (on IBM zSeries) to the SPEED server. The first set at beginning of day (BOD) had 16 files giving the complete position data. Subsequent sets of files were transferred every two hours from the core systems to SPEED, carrying only updates. These files were manually copied into a shared directory on a file server.

    Later, the system was modified to allow a human operator to upload the files using a browser-based file-upload interface on the SPEED system. This made file uploads easier and shortened the total time taken to process a batch.

    In the third version, the files were pushed out through a shared file server drive automatically, through scheduled batch jobs on the core system. Corresponding batch jobs were fired by the scheduler on the SPEED system to pick up recent files from the incoming directory and process them. This allowed almost full automation in data uploads.

    When SPEED was first deployed, the BOD batch was uploaded at 9 AM and subsequent incremental batches were uploaded every two hours till 6 PM. Over the next three years, this was steadily increased till the first BOD batch was uploaded at 3 AM and subsequent updates were uploaded every two hours till 8 PM. At this point, the fully automated scheduler-driven generation and upload of batch files was necessary, since this duration and frequency was not feasible with manual uploads.

    There were monitoring programs on the SPEED server to check correctness and completeness of each batch. Both success and failure were reported through emails and SMS messages generated by the SPEED server.

    Strategic Significance This project was the first system where NSDL published dynamic business data on the Internet. This was also the first system where NSDL used Linux in a production system. We were given the privilege to partner with NSDL to cross these two milestones. The system was designed and built in about three months in early 1999, and was decommissioned in early 2004 when a larger online system was built to replace it.