7 FEB, 2012, 02.11AM IST, JAYADEVAN PK,ET BUREAU
BANGALORE: In a Bangalore suburb that is its nerve centre, the Aadhaar project, which aims to give over 1.2 billion Indians a unique ID number, is proving in many respects to be another example of a typically Indian trait - jugaad.
The network operations centre for the world's largest exercise in biometric identification is housed in a nondescript rented building. There are no supercomputers, just low-cost commodity hardware for computing and storage. And the technology centre for Aadhaar employs just 250 staff.
But beneath the frugality that gives the appearance of a rough exterior lies ingenuity and cutting-edge technology that enable a million Aadhaar numbers to be generated each day.
"Our operations are more similar to a Google or a Facebook than to anything else," says Srikant Nadhamuni, who helped design the Pentium chip for Intel, and is the head of technology for the Unique Identification Authority of India (UIDAI).
Aadhaar, it is expected, will not only plug leakages in the public distribution system and subsidy schemes that cost the government nearly 3 lakh crore every year but also serve as the single most important source of identity for all Indians. Nadhamuni says because of the enormous technology challenges of speed and scale, Aadhaar is an object of attention. "Anywhere we go, we are the talk of the town," says Nadhamuni.
Surrounding the screens providing a running commentary of Aadhaar's progress are workers in cubicles making sure that all the systems associated with the project are working fine. If there is a snag, they are the first ones to be alerted.
Aadhaar equals US biometric database
Last month, the screens showed that the Aadhaar number count was inching close to 120 million, about a fifth of the way to its target by 2014.
In just over a year-and-a-half since the journey began, Aadhaar has equalled the size of the world's largest biometric database maintained by the United States. Perhaps the most important piece of the Aadhaar project they keep an eye on is what is called 'de-duplication' to make sure that no person gets more than one identity number. Each record is compared against all the existing records in the database.
Nearly 100 trillion biometric matches have to be carried out every day to deliver one million Aadhaar numbers. If a duplicate slips through, the vendor is penalised, says Pramod Varma, chief architect of the UID project.
This has been made possible because the UID authority appointed three vendors and made clear to them that the proportion of work allocated to them falls every time they fail to identify a duplicate. Vendor systems working on the data centres are put on separate networks guarded by firewalls and isolated.
After de-duplication, the data packets are archived, encrypted and stored on a separate system. The data is divided into parts and stored in different databases so that no single one holds all the data.
The data centres will eventually contain 15-20 peta bytes, or about the equivalent of 20 million copies of the Hindi movie Sholay.
Access is controlled using biometrics and multiple levels of physical and logical security. "It has perimeter security and (is) monitored using cameras," Varma says. Inside the data centre, there are zones with different levels of security. Firewalls procured from multiple vendors guard the equipment inside.
Recently, the UID authority tested nearly 84 million records and found that 99.86% of the population can be uniquely identified by the biometric system. They also found that 99.94% of all duplicate entries have been caught. The initial cost of de-duplicating a record was over 20 per record but it was brought down to 2.75 because of multiple vendors competing for a bigger slice of the business.
Rishikesha Krishnan of IIM-Bangalore is of the view that the project has allowed India to demonstrate its IT prowess on a scale never seen before. "It has proved the ability to scale up in a manner comparable to the Delhi Metro." The project is also characterised by open standards and open source for software development, says Varma, who was part of the founding team of supply chain solutions provider Yantra Corporation that is now part of IBM.
Most of the technology used is obtained off the shelf and an open application programming interface works on different platforms, including Linux. All the data is stored in two centres which occupy 5,000 sq feet. Nadhamuni and his team want the data centre to be a 'lights out' centre, where a human will never have to enter.
"If the UID had tried to build everything and had not been open, it would have been impossible. It would have cost 50 times more than Infosys' strength (of over 1 lakh staff)," says Nishant Shah of the Centre for Internet and Society.
Before de-duplication, the data packet begins its journey at the enrolment centres where a person's fingerprints and iris scans are taken using standard software and hardware. The information is validated by a government official and is checked for quality by the enrolment software.
At the end of the day, data is reviewed and signed off by a registered supervisor and sent to the UID authority. But before the data is saved, it is encrypted. When the data reaches the central server, a random manual quality check is done for the photograph, gender and age.
Nandan Nilekani, the Infosys co-founder who is the prime mover behind the project, says Aadhaar is on track to enroll 60% of India's population by 2014. "It is now the largest, fastest and the most accurate biometric database in the world."
