Net neutrality and Internet QoS in Sri Lanka redux

Lirneasia’s post on Net Neutrality by Chanuka Wattegama stimulated a lot of debate on the pros and cons of net neutrality from a Sri Lankan perspective. In Net Neutrality: Economics and implications for ICT4Peace and ODR I fleshed out some of the implications of Net Neutrality, and Chanuka’s stance on it, for web and internet traffic related to peacebuilding and Online Dispute Resolution. In it I also critiqued Chanuka’s assertion that ” that heavy users should pay for the additional bandwidth they use” would in any way help address Quality of Service (QoS) issues stemming from the lack of international bandwidth (or more accurately, the high cost associated with better bandwidth) in Sri Lanka.

My post prompted the head honcho of Lirneasia to suggest that it really had no position on Net Neutrality. My response to that assertion was followed by another post by Chanuka on Lirneasia’s site, suggesting the debate on net neutrality was of enduring interest in Sri Lanka.

Like Plusnet in the UK, Comcast (in response to the FCC’s ruling) has now come out with what seems to be a sensible, protocol agnostic, traffic management plan. Though different to Vint Cerf’s model of managing traffic on the internet, the Comcast plan looks very interesting on paper. As Ars Technica notes,

Comcast’s new technique is based on a simple premise: during periods of congestion, heavy users of bandwidth on a local node ought to see speed reductions before light users. To make that happen, the system tracks each customer’s uploads and downloads separately using software from Sandvine that runs on Linux servers (Comcast stresses to us that this is not deep packet inspection software, but basic “shallow inspection” code that simply counts packets.)

When any port (think neighborhood node) on the Cable Modem Termination System (CMTS) in the local cable company office enters a “near congestion” state, the system looks up the heaviest users of bandwidth during the preceding few minutes. Those users then have their traffic tagged as “Best Effort” rather than the default “Priority Best Effort.” At this point, nothing happens to anyone’s traffic.

When congestion actually occurs, the Priority Best Effort users should see no slowdown in their connections; all traffic will go through ahead of the Best Effort traffic. Best Effort folks may not notice any slowdown, either. They are not speed-limited, but they do go to the back of the quality of service (QoS) line. At this point, if traffic does in fact fill the pipe, users in the Best Effort category will experience delays in their connections, though their traffic will still be sent on whenever possible.

Emphasis mine. 

Deep packet inspection based traffic management is a route that spells disaster for freedom of expression on the web, particularly under repressive regimes that under the guise of say protecting citizenry from pornography can institute a strict regime of over-broad web filtering that censors inconvenient truths. 

Deane’s assertion on Lirneasia’s blog, that the market will look after itself without any regulation and that net neutrality is “American-progressive worry about the free market”, is ignorant of Lirneasia’s significant research on Internet QoS and how (Sri Lankan) telecoms companies will act against the interests of consumers, as noted by Janaka Beneragama and the comment here (though Janaka’s conflation of international bandwidth and unlimited downloads is just wrong).

Further, Lirneasia’s own field-testing of 3G “broadband” points to a glaring divide between marketing hype and reality.

In this context, Sri Lanka Telecom (SLT) on 12th September 2008, under a misleading PR titled SLTnet goes 3G, stated that, 

SLTnet, the internet arm of SLT which is the largest internet service provider (ISP) in Sri Lanka today offers up to 3 Giga bits per second (Gbps) international internet bandwidth, adding even more capacity to be of better serve to the nation… With this initiative, SLT provides internet users in Sri Lanka super fast access to online web applications such as web search, web mail, calendar, images, VoIP calling applications, audio, video and maps amongst a host of other most popular features on the internet…  SLT is happy to invite and facilitate the hosting of mirror sites of all popular international sites like Google, Yahoo, MSN, Facebook, Youtube etc., to improve the quality of services to Sri Lankan internet users and to help avoid the bottlenecks found in the international internet backbone.

I wonder what the policy analysts at Lirneasia would make of this announcement? On face value, it seems to respond to some of the key findings of the Ashoka-Tissa QoS survey. As noted by Chanuka, “What we saw from our research that limits in international bandwidth is the key reason for the poor broadband experiences in India and Sri Lanka.”  But the PR also has gems like this,

Also SLT is now in the process of further upgrading and improving the direct connectivity with social networking, web application and video service providers as and when the need for broader bandwidth arises.

which is utter nonsense. Let’s also not forget that two years ago, SLT promised us VDSL with speeds of 52Mbps.

In Patriotism and broadband in Sri Lanka I explored a similar announcement by Lanka Bell to strengthen its internet backbone. So in sum Lanka Bell has a 1.2 Terabit cable backbone and SLT now has a 3 Gigabits per second backbone. Dialog I believe has its own backbone.

This sounds like an awful lot of capacity. Yet, to date, my ADSL Office Express connection from SLT, however, shows absolutely no signs of improvement. Today, as on most Sundays, I got a maximum sustained download rate of 224.5kbps and a peak upload rate of 103kbps on a connection that promises much more and for which, may I add, I pay a premium. Tomorrow, after 8am and especially around 5pm, I know my transfer rates will be no better than my erstwhile dial-up modem. So where’s the real benefit to the consumer beyond the marketing spiel?

The issue of net neutrality for Sri Lanka is quite simply this. Sri Lankan telcos will, before investing on local infrastructure and international bandwidth, always default to traffic management to make do with what they already have across their customer base. Improvements to QoS will be infrequent. Being a small market (our entire broadband market is is dwarfed by broadband consumers in large Indian cities alone) yet one that increasingly creates, disseminates and accesses audio-visual content on and for the web, the demands placed on local ISPs to guarantee minimum data transfer rates will continue to grow. However, ISPs will continue to only guarantee maximum data transfer rates, capping and managing as they see fit our use of their pipes. Verbose FUPs will attempt to convince us that network management works in our favour, when the reality will be quite different. And whatever promises SLT, Dialog Telekom and Lanka Bell make on enhanced access to the web, we know that we’ll never be able to access Tamilnet through their pipes. 

While I look forward to mirror sites and the benefits they will bring to long-suffering consumers, I also wonder if, in the absence of progressive regulation, telcos that obey every arbitrary diktat of the Ministry of Defence and under a TRC more interested in appeasing the Rajapakse regime’s lunacy, these seemingly progressive measures will be used to control and curtail our behaviour on and access to the web. 

For me net neutrality is more than serious concerns about deep packet inspection. It is about the commitment to an open web, where regulation is more transparent and progressive than partisan, stentorian and limiting, where government promotes access to the web and Internet as a right of all citizens, where everyone is guaranteed a minimum QoS and where select content isn’t discriminated against. 

I guess given a choice between slow access to an unrestricted internet over blazing fast access to a restricted internet, I’d gladly choose the former.

But is asking for both really that unfair?

UPDATE – 27 September 2008

Lirneasia’s head honcho the good Prof. Samarajiva and I have an interesting debate on the issue of net neutrality starting with his response to this post here.

Net Neutrality: Economics and implications for ICT4Peace and ODR

A post on Lirneasia prompted some thought on the linkages between Net Neutrality and peacebuilding, especially the use of the web and Internet for conflict transformation. Lirneasia’s post deals with Obama’s and McCain’s stance on the issue of Net Neutrality, with Chanuka making the point that while theoretically desirable, Net Neutrality has its own significant costs.

A complementary article posted earlier on Lirneasia’s site itself points to an approach by Vint Cerf that provides useful food for thought on the Net Neutrality debate. Cerf’s agrees that broadband networks need to be managed, but he differs with Chanuka (and perhaps Lirneasia) on how. As opposed to usage based billing, Cerf proposes a transmission rate cap where users can “purchase access to the Internet at a given minimum data rate and be free to transfer data at at least up to that rate in any way they wish.” (Cerf’s original post on Google which fleshes this idea out can be read here). 

My concern here is with the appropriation of the Net Neutrality debate by ISPs – both State and Private – under repressive regimes to covertly clamp down on communications used by human rights defenders and peace activists. 

For example, I have been reliably told, though not verified, that a well-known ISP in Sri Lanka (not SLT) is blocking P2P traffic, including Skype. This creates significant problems for some HR org’s and activists on it who use Skype to communicate and collaborate securely. Ironically, some actually switched over to this ISP from SLT because they thought it afforded greater security and Quality of Service. EFF’s Switzerland tool, if Lirneasia or any other organisation ever get around to using it in SL, may offer some insight in this regard.

The point is quite simply this – net neutrality is not just about the minimum or maximum transmission rates, but about the way IP packets on a broadband pipe are managed. If ISPs, under their own misguided policies or those covertly imposed by a repressive regime begin to selectively prioritise and monitor traffic on their networks, it forces those who use the Internet for highly sensitive communications and advocacy to re-think the tools and services they access, and how. And sometimes, there’s no other option for tools used by HR defenders – as in the case of Skype. Despite recent concerns over privacy, there is no other encrypted, free and widely used VOIP tool. And once you start going down this path, it soon becomes clear that traffic discrimination can selectively target other tools, web services and platforms used by HR defenders against a regime to capture, generate, disseminate and archive inconvenient truths – such as human rights abuses. This includes video streaming sites like YouTube.

A final word on economics. As Ars Technica notes,

As unattended apps like P2P and network backup utilities tie up a portion of bandwidth for ever longer periods of time, the old solutions aren’t working as well and congestion is one result. Cerf’s idea would take us back to the old “circuit-switched” days in the sense that each Internet user would instead get a guaranteed line with a minimum guaranteed rate at all times. This would answer consumer complaints about “not getting what I paid for,” but would cost ISPs more cash.

Emphasis mine. Lirneasia’s research in Sri Lanka suggest deplorable QoS across all “broadband” ISPs. Not a single ISP in Sri Lanka guarantees minimum transmission speeds and often advertise speeds that paying customers simply don’t get, or even come close to. Convincing them to upgrade their networks to go down the path Cerf suggest may be impossible, given how enticing the economics of a metered data transmission model looks and sounds, on paper. 

The problem of course is that this doesn’t address the problem of pissant data rates for all. A pay-for-megabyte model will see that though the heaviest users pay up (corporate consumers) and the economic disincentive for individuals to become high volume users will simply not be enough to improve transmission speeds (particularly if, as I suspect, our ISPs will do little or nothing to improve network capacity). The net result will quite simply be more or less the same old, glacial data transfer rates which will anger even more those who can are willing to pay more (like myself) for better connectivity. 

There’s one ISP in the UK offering something I’ve not seen anywhere else – a meaningful IP traffic prioritisation / management plan. It’s from Plusnet. Check it out here. Their explanation uses the same metaphor as Chanuka uses in his Lirneasia post,

Think of it this way, the broadband network is like a motorway. When the traffic is light, all vehicles can move at the national speed-limit. Some lanes of the motorway have been reserved for important traffic, such as buses or emergency vehicles. During rush hour, most vehicles are forced to slow down. However, the traffic on the reserved lanes can continue to travel at their full speed.

Google itself has promised a tool that helps end-users / consumers to see how ISPs manage traffic. No date for the release of the tool, but a more user friendly Switzerland or Google’s tool would be a huge asset for those of us who use the Internet for peacebuilding and ODR, if only to see which ISP we should avoid.

Update – 5 September 2008

Comcast, the cable operator and ISP in the US at the centre of the Net Neutrality debate, has sued the FCC over a decision it made on Comcast’s network management techniques. Ars Technica has the story here.