A deal over the dark fibre

Is U.S.-based IUNet riding the back of education, which is in dire need of high-speed connectivity, to commercial ends?

A LOT has been written and said about the Sankhya Vahini project, the proposed joint venture between IUNet Inc., a wholly owned subsidiary of the Carnegie Mellon University (CMU), Pittsburgh, United States, and the Department of Telecommunication (DoT)/t he Department of Telecom Services (DTS) of the Government of India (GOI) to set up a high-speed data network over optical fibre, exclusively for data transmission and no voice, as the country's high bandwidth Internet backbone, which has now got mired in a great deal of controversy after the Cabinet gave its approval on January 19. A variety of issues have been raised at various levels - political, parliamentary, commercial, bureaucratic and technical - and there have even been demands from various quar ters to scrap the project altogether.

The government, on its part, has sought to defend the project and clarify these issues, though not entirely satisfactorily.

The Joint Venture Agreement (JVA) to form an Indian company called Sankhya Vahini India Ltd. (SVIL), which was initiated on August 28, 1999, based on the Memorandum of Understanding (MoU) signed on October 19, 1998, and the basic Technical and Business Plan (T&BP) jointly arrived at in January 1999, remains to be signed. That would depend on the final go ahead from the government.

The pronouncements of Communications Minister Ram Vilas Paswan on May 15, that of the Human Resource Development Minister Murli Manohar Joshi earlier this month, and that of Information Technology Minister Pramod Mahajan on April 15 - the three Ministrie s that will have an equity holding of 45 per cent, 4 per cent and 2 per cent respectively in SVIL - that the project will not be scrapped, suggests that the green signal is imminent. Significantly, it was Pramod Mahajan who had fuelled a good deal of the controversy.

Notwithstanding the above, there are still many questions that have not been adequately addressed despite the long-drawn controversy. The data security angle has been blown out of proportion; actually it is a non-issue, according to experts. Technicaliti es relating to the legality of IUNet (which was incorporated in Pennsylvania on January 8, 1999) as an entity at the time of signing of the MoU, its true relationship with the CMU and the issue of global tendering for the technology, among other things, are not serious and can perhaps be satisfactorily explained. Since there is still time before the JVA is signed and the company in India is incorporated, it is important that the more serious issues are taken care of in the interests of the stated long-t erm objectives of the project.

The genesis of the project is significant. According to the document issued by the government in April in the wake of the controversy, the project was "conceived" during the National IT Task Force meeting of September 5, 1998, in New Delhi wherein the Do T - now rechristened DTS - was made a nodal agency to work with IUNet (short for inter-university network).

The meeting was chaired by Andhra Pradesh Chief Minister N. Chandrababu Naidu, who co-chaired the Task Force. Much has been read into Chandrababu Naidu's involvement and push to the project - specifically his close association with the non-resident scien tist from Andhra Pradesh, Dr. Raj Reddy, who is one of the brains behind IUNet and its chairman. Raj Reddy is a Professor of Computer Science and Robotics, the Dean of the School of Computer Sciences at CMU and the co-chair of President Bill Clinton's IT Advisory Committee. While this certainly must have been an important factor in Chandrababu Naidu's extraordinary interest in the project, it does not in any way lower the credentials of Raj Reddy, the winner of the Alan Turing Award, the highest award f or computing sciences. Similarly, the long-standing association of the other brain behind IUNet, Dr.V.S. Arunachalam, former Scientific Adviser to the Defence Minister and at present a distinguished services professor at the Department of Engineering and Public Policy at CMU, alongwith M.G.K. Menon, the other Co-Chairman of the Task Force, is believed to have played a significant role in recommending the project as a joint venture with IUNet.

The decision was formally approved at the November 21, 1998, meeting of the Task Force, which was also chaired by Chandrababu Naidu. Interestingly, the minutes of the September 5 meeting in which a presentation of the concept had apparently been made to the Task Force, do not seem to have been circulated to the Task Force members, a point that has been subsequently raised by P.V. Jayakrishnan, the present Secretary in the Ministry of Information Technology (MIT), formerly the Department of Electronics ( DoE), in his letter of April 18 to the Telecom Commission. This has been corroborated by other members of the Task Force. According to them, in keeping with the characteristic style in which the Task Force meeting was conducted, the project was pushed th rough as one of the many items on the agenda without any discussion even as unbiased members raised some questions.

Irrespective of the merits of the project and the credentials of IUNet and its promoters, all evidence points to the fact that the administrative decision to enter into a joint venture with IUNet was made in an irregular manner. As a member of the Teleco m Commission, Jayakrishnan has also pointed out that the Telecom Commission was merely informed on November 11, 1998, that an MoU had been signed between DoT/DTS and IUNet but a detailed deliberation on the subject, which the Commission is mandated to un dertake on the introduction of any new telecom service, was never held. This does not amount to the Commission's approval, he has pointed out. Apparently, the draft T&BP of SVIL was never submitted to the Commission as had been promised by the DoT/DTS. O n this issue, the government's reply is at best evasive.

A pertinent point has been raised by Nilotpal Basu, the Rajya Sabha member of the Communist Party of India (Marxist), in his letter of May 1 to the Prime Minister. While expressing concern over the project, he has pointed out that the terms of reference vested only an advisory capacity to the Task Force and not any executive authority to decide on specific projects and "even less specifying the joint venture partner or evolving financial parameters". The Parliamentary Standing Committee on Communication s, headed by the CPI(M) MP Somnath Chatterjee, in its latest report has also rightly questioned the Task Force's authority to decide on a particular joint venture partner without the requisite expertise or a study to evaluate its technology against simil ar technologies and other competitors. Given that there is no evidence that any executive body of the government had processed the Task Force's recommendation and taken any appropriate decision, it is indeed intriguing on whose directive the DoT signed t he MoU.

Basu has called for a public enquiry into this aspect. Considering that the Task Force had the blessings of the Prime Minister's Office (PMO) all through, Sankhya Vahini too had the political backing what with the imperative of the Telugu Desam's support for the Bharatiya Janata Party to stay in power.

BUT purely on technical grounds, one of the key reasons for the Task Force even considering the idea of setting up a high-speed data network with a focus on educational institutions was the failure of the DoE to expand and upgrade the existing network fo r the educational and research community, Ernet, into a high-speed/high bandwidth network linking all universities and research institutions. This high bandwidth network, called the S&TNet, never got off the ground even though a funding of about Rs.100 c rores had been earmarked for it, with about half of it to be met from extra-budgetary resources (such as Ernet's earnings). The DoT had, in fact, agreed to lease 34 mega bits per second (Mbps) bandwidth for the purpose (as distinct from releasing the ent ire fibre as is being proposed for SVIL), the DoE never got its act together and the budgetary allocation made till date (about Rs.60 crores) has been spent in just tinkering with Ernet's existing services with no visible improvement at the user-end.

As a result, the educational and research community, which is in dire need of high-speed connectivity, was getting frustrated. Besides, there was a growing need for bandwidth for other applications in the IT sector: the software industry, Internet users, Internet Service Providers (ISPs), distance education, IT enabled services and the imminent growth of e-commerce.

Arunachalam had apparently been talking about this concept of CMU partnering in India to develop a high-speed backbone network at various forums, including his earlier presentations to Prime Minister P.V. Narasimha Rao in 1994 and to the Telecom Commissi on in 1997. On an invitation from the Chairman of the IT Task Force, Jaswant Singh, he and Raj Reddy made the presentation at the September 1998 meeting of the Task Force. This set the grounding for the Task Force, mandated as it was to recommend solutio ns for the national information infrastructure (with 2.5 giga bits per second backbone), recommending a joint venture of IUNet with DoT. Given the DoT's existing 2.5 Gbps capacity 10,000 route-km of single mode optical fibre, lying unused for long, the v iew that DoT by itself would not be able to implement such a high-speed backbone quickly is likely to have influenced the Task Force's recommendations.

According to Arunachalam, separate discussions with Jaswant Singh and with Sushma Swaraj, then Minister of Communications, and where Chairman, Telecom Commission, Member (Services), Telecom Commission, and Convener, Task Force, were present, led to an in vitation by the two to participate in a joint venture.

Arunachalam and Raj Reddy agreed, provided a few leading Indian educational institutions would also partner with DoT and CMU in the venture.

The educational and research community can at best use up about 16 to 32 Mbps bandwidth, according to experts who have monitored Internet usage in the educational sector in India. A cost-benefit analysis shows that the use of a single-mode fibre is best optimised at 2.5 Gbps bandwidth. Nearly a hundred-fold capacity would, therefore, be available for other applications, mainly commercial. From the perspective of DoT/DTS, therefore, there was an enormous revenue earning potential visible in this joint ve nture by speedy "lighting" of these "dark" fibres for carrying Internet Protocol (IP). The current policy does not allow leasing of this fibre to any private party; nor is the laying of such a backbone by an ISP an easy proposition in the Indian context.

Given DoT's available infrastructure, this headstart in providing Internet access implies a virtual monopoly for DoT/DTS. Also from the perspective of IUNet, which, notwithstanding its affiliation to a university, is a company whose bottomline is profit. One cannot expect altruism here, and commercial prospects would override all other concerns as is indeed evident in the JVA. Once the needs of the educational institutions are taken care of (but even this objective is unlikely to be fulfilled) the huge remaining bandwidth has to be leased to business applications, including long distance educational ventures by a large number of foreign institutions and companies, which are just waiting for the bandwidth to be released. The affiliation to CMU and the o stensible objective of providing high-speed Internet access to the educational sector would seem to be only a convenient premise to launch into business with the IP.

And a 49 per cent equity share without investment in cash - equity being valued against the equipment to be supplied by IUNet - implies a corresponding share in the profits. The T&BP envisages profits from the fourth year onwards. A Rs.26-crore profit in the fourth year is estimated, which would increase to Rs.555 crores (post tax) against a total capitalisation of Rs.1,407 crores at the end of the tenth year.

THE general opinion in the educational/research community as well as business circles, and even within MIT, is that left to the government, such a network will never get established. Hence ways have to be found to put this infrastructure in place if the IT based education, research and industry must grow. The moot question is whether a joint venture with a foreign company is the right route and, if yes, whether IUNet is the best choice from the spectrum of technologies and vendors available for high ban dwidth networking solutions. Here one must distinguish between a joint venture with, or a turn-key job by, a multinational telecom company like AT&T. While Raj Reddy and Arunachalam themselves may not be experts in optical switching technologies and high -speed networking, being in a university set-up, the expertise of CMU specialists was available in the designing of the Sankhya Vahini network. More important, Indian network specialists from the DTS, the Indian Institute of Science (IISc) and the Indian Institutes of Technology (IITs) were also part of the design team, and the final design was a joint effort, something which will not happen with a multinational company.

However, one must add that there is nothing extraordinarily innovative or original in the structure of Sankhya Vahini. According to IISc experts, if we put together a bunch of experts in India, with some resources, somewhat similar architecture would hav e resulted. The difference is that the experience of CMU specialists gained from designing and running such networks over a period of time - aspects such as where to put add/drop multiplexers (ADMs), the router switching protocols, how to space the optic al signal repeaters, and use of newer concepts like Dense Wavelength Division Multiplexing (DWDM) (which Sankhya Vahini is designed to incorporate in order to enhance bandwidth) - would be brought to bear upon the network solutions. According to Arunach alam, however, until their presentation of the data-only concept to the Task Force, there have been no reports of other organisations in India (or elsewhere) coming forward with the setting up of such a network. "The opportunity through Sankhya Vahini is to leapfrog in technology, based on IUNet's design, incorporating cutting edge technology and equipment (for which there will be no royalties or technology transfer fees to be paid). Most global service operators in the world today operate different net works, based on legacy equipment, and so their experience is different. Sankhya Vahini will be a gigabit network available to the user, a major step for any network today." (The government has, however, falsely stated - for whatever reason - that "the pr imary patent on DWDM technology... is held by CMU". This is simply not true and DWDM is offered by many firms today.)

A probable positive side to the venture is easy access to certain technologies otherwise restricted for export from the U.S. Optical-switching technologies, gigabit routers and repeaters for high bandwidth are controlled technologies.

Apparently IUNet has given an assurance that these technologies would be accessible to a university, and for collaboration in the research and education sector these should be exportable as well. But this is not obvious because the transferring entity is not a university but a company, and the usual export control laws of the U.S. Department of Commerce would apply. More so, because the network will be accessible to all organisations, including those on the Entities List facing U.S. sanctions in the wak e of Pokhran II. Export restrictions may well be the reason why the fibres have remained "dark" and naturally, DoT/DTS jumped at the opportunity provided by the joint venture.

In many ways, the approach adopted here to IP traffic solutions is similar to what Sam Pitroda, as a switching expert, brought to telephony in the country. If C-DoT (Centre for Development of Telematics) attempted to circumvent the entrenched bureaucracy of DoT, SVIL should do the same for the Internet. But there is an essential difference. Pitroda became part of the government, transferred technologies and worked from within with no commercial profit motives. Arunachalam and Raj Reddy, on the other han d, are collaborating as businessmen, with clear profit motives, albeit with some academic leanings and possible access to technologies.

HOWEVER, the overall impression that has been sought to be conveyed is that the proposed network is primarily intended for the educational sector, one of whose promoters is a U.S. university subsidiary established with the specific charter to design, dev elop and manage high-speed data networks internationally for educational purposes. The name IUNet is coined to convey the same. The envisaged participation from Indian educational institutions and universities is meant to achieve this ostensible purpose of involving the skilled manpower pool in the development, evolution and use of advanced networking technologies.

However, a closer look at the JVA belies the stated objective in many respects, and education - as public good as distinct from private enterprise - would be least served by this network. Indeed, besides the high-profile institutions such as the IITs, th e IISc and the Indian Institute of Information Technology (IIIT) and the university system, which is supposed to be one of the major beneficiaries of the network, was at no stage a part of the discussions leading to the design of the network. It would se em that the intention here is to ride the back of "education" - public good - to commercial ends. Consider for example the fact that the IITs, the IISc and the IIIT and other educational institutions under Ministry of Human Resource Development (MHRD) (w hich would include the universities), which together will hold an equity share of 4 per cent, have to subscribe to their share of equity. For the initial paid up capital of Rs.300 crores, this amounts to Rs.12 crores. First, no institution has that kind of money and, more pertinently, the statutes of these institutions do not allow this.

Apparently, during the earlier stages of the negotiations, representatives from the IITs and the IISc had proposed that considering that they would be contributing in terms of knowledge generation, sweat equity should be allowed for educational instituti ons and the total equity share should be higher than 4 per cent. Indeed, these institutions have already contributed to the network design.

Institutions can accept sweat equity shares just as they accept technology transfer fees. But both the pleas have been ignored. In all likelihood, with the educational institutions unable to subscribe, their share of equity will be gobbled up by DoT/DTS or MIT. Only half of this equity is available to DoT/DTS or MIT. The other 2 per cent will have to be off-loaded to some other non-governmental parties.

According to Arunachalam, they had insisted at the Task Force meeting that educational institutions should be involved as equity holders in the project. "We pointed out that Indian universities should be invited as U.S. universities regularly participate in network and IT projects. We were conscious that this would be a revolutionary step as Indian universities have not as yet formed commercial ventures to capitalise on their intellectual assets. We pleaded with DoT that these institutions should be giv en sweat equity, and more educational institutions be included as appropriate. As the government had not yet worked out plans for sweat equity, this suggestion was over-ruled," he says.

Representatives of Indian institutions had also argued for the revenue earnings of the network to underwrite the expenses of universities and educational institutions. But the pricing policy laid down for access to the network is uniform, irrespective of the user. That is, there is no cross-subsidy that one would expect for the cause of public good.

Moreover, universities and colleges have to invest in equipment such as high-end personal computers at the user-end. Universities and colleges are as such starved of funds. Will the MHRD allocate money specifically for this purpose? Indeed, there is no m andatory requirement on Sankhya Vahini to reserve a certain fraction of the bandwidth for public good.

Arunachalam says: "Unfortunately, today, the Indian cost of bandwidth is many times higher, and that too for poorer quality (oversubscribed) and unreliable bandwidth. This is especially problematic for educational institutions as they are traditionally p ioneers in the use of high bandwidth. There are two options for making such bandwidth available and affordable to them. One, is to cross-subsidise by charging commercial users extra tariff, or look for educational subsidies from the government. If one we re to build a network solely for educational users, it would not be cost-effective given the economies of scale with newer technologies that demand more users to keep costs low. A practical option should therefore involve making a scalable, high-bandwidt h network for all users, bringing down the costs across the spectrum. This option is realised by developing a data-only architecture with very high bandwidth capability." He argues: "This does not preclude Sankhya Vahini from having special pricing for e ducational institutions, as well. However, even the average project tariff will be much lower than what educational institutions pay today for much lower bandwidth. We have, in fact, modelled exclusive bandwidth availability for educational purposes, at a level found only in a handful of countries."

This argument not withstanding, it would appear that DoT / DTS have failed to put their minds to draw up an equitous JVA. For example, if IUNet's equity is measured in terms of services offered, Indian institutions should also be entitled to equity for t heir intellectual input and services. Indeed, the JVA places the Indian entities and IUNet on an unequal footing in various respects. For instance, if collaborative R&D with CMU is envisaged as is being touted, while the SVIL Board can have three represe ntatives from CMU, the Indian educational sector can have only one. In operationalising and exercising control over SVNet, IUNet rather than the Indian company appears to be the prominent player. The entire responsibility of identifying equipment and tec hnologies has been entrusted to IUNet whereas, as pointed out by Jayakrishnan in his note to the Telecom Commission, it would be more appropriate for the SVIL Board to identify equipment and technologies to be sourced and, if necessary, the same may be c ontracted out to IUNet against which equity would be adjusted and the rest of IUNet's share invested in cash. Also, the JVA constantly refers to IUNet "selling" equipment. Actually, IUNet will merely be a facilitator, or a middlemen, sourcing equipment f or SVIL.

There is a curious element in the JVA, which defies an immediate explanation. The agreement repeatedly refers to the fact that at no stage can the government's equity in SVIL at no stage can exceed 49 per cent! This seems to be privatisation (or disinves tment) with a vengeance from the word go. Even if the venture were to become profitable, the government (DoT/DTS and MIT together) - educational institutions do not constitute government - cannot raise their equity to more than 49 per cent. This is why o nly 2 per cent of educational institutions' share of 4 per cent can be picked up by the government.

One can understand a ceiling on IUNet's equity in SVIL but not a ceiling on the government's equity in an Indian company. From the minutes of the Telecom Commission's meeting of November 11, 1998, it seems that it was the Commission's directive that equi ty of the government should not exceed 50 per cent. (For the Commission to issue this directive without ostensibly having discussed or understood the details of the various aspects of the joint venture, seems curious, even contradictory.) Now the questio n is whether such a directive was given on insistence from IUNet or is it a reflection of some policy decision. The government should clarify this.

On May 16, Paswan said that it would be made conditional that after the five-year lock-in period, IUNet would be allowed to offload its equity only to Indian investors. Now CMU, which at present has 100 per cent share in IUNet, will have a majority share only for a five-year period, after which CMU can off-load its shares in the market which can be picked up by any entity, including MNCs like AT&T. This could mean that controlling interests in SVIL would be transferred to a company with no links whatsoe ver with the education and research sector. The question is whether a caveat against this has been built into the JVA. It may well be that, in the light of the above, joint venture is the best route to gain access to technologies for implementing a high- speed data backbone for India. But there are a lot of questions that remain to be addressed. A committee should be constituted to go into the merits of the joint venture and give its recommendation. A couple of months delay is not going to matter. If the JVA is rewritten so that SVIL reflects Indian interests better - particularly elements of public good - and the various apprehensions sorted out, there can be true light in the dark fibre.