South Asian Broadband Service Quality: Diagnosing the Bottlenecks

The developing countries have lately experienced a surge in mobile phone adoption. New investment into the telecom sector has enabled network roll-out, and increasingly innovative business models have driven down network operation costs and customer acquisition & retention costs, enabling pricing to be lowered significantly (Samrajiva, 2009). The result is that people who, just a few years ago were unable to afford any form of telephone are now purchasing mobile phones and SIM cards and using them to make calls and send SMS. Given the dearth of fixed access networks, developing nations, especially those in South Asia, depend mobile phones to take their citizens online. All these new consumers are naturally creating increased demand for Internet capacity. At the same time, the demand for Internet access capacity is increasing because many South Asian countries (particularly India, and to a less extent others such as Sri Lanka) have been at the forefront of attracting a significant share of the booming market for business process outsourcing. While new consumer numbers are growing, tests performed by LIRNEasia in India, Bangladesh and Sri Lanka show that users get low value for money from their broadband connections when compared to North American counterparts. Using only data publically available, we could hypothesize that the poor performance of South Asian broadband in terms of throughput (upload speed, download speed), jitter, latency and other measurable dimensions is due to the lack of international capacity (as opposed to the local access network capacity or in-country back haul capacity). For example, we know that while the region has surging demand, the supply of international connectivity has not caught up to the demand (Telegeography, 2010). Therefore South Asian internet service providers end up paying significantly higher fees for international connectivity when compared even to their East Asian peers (Telegeography, 2009). We test our hypothesis using a broadband quality of service testing methodology developed jointly by LIRNEasia and the Indian Institute of Technology, Madras. The results confirm that international connectivity is indeed a significant choke point when accessing the Internet and negatively impacts the service quality of broadband. Individual countries or individual ISPs cannot do much in the immediate or near term to increase international leased line capacity and prices (due to relatively long lead times required to actually lay the cable, and the seemingly longer time some countries require to agree on the rules related to the consortia that participate in the cable). However, our testing provides data on a second choke point, one that is very much within the control of the ISPs. The points where two or more ISPs exchange traffic (network access points/NAPs, or Internet exchanges/IXPs) turn out to be choking points also. While ISPs in smaller/developing countries may not have control over international NAPs/peering points, the local NAPs and IXPs are very much within their control. Real-world examples show that when a country sets up at least one IXP, it is able to achieve a significant rationalization of demand for international capacity because local traffic is kept local, instead of being exchanged internationally. The resulting lower cost and higher quality of the connectivity justifies the relatively small investment required to set up an IXP. However due to a host of issues, it is possible to have an IXP and not get the benefits – we show quality testing data from India (which has a network of IXPs) that confirms this. Therefore while we push strongly for the establishment of IXPs, we will caution about getting the technical models and business incentives aligned towards its proper functioning.