Local Loop

It was inevitable that I would eventually post something about the punny namesake of this blog: the “local loop” in South Africa. This is the last-mile (copper) infrastructure owned and controlled by the incumbent telco: Telkom SA

Local loop unbundling (LLU) is a regulatory policy that forces an operator with “significant market power” to allow competitors to directly resell their access network. Predictably, the incumbent operators always push back against LLU, and this gives rise to different “flavours”. “Full LLU” means access to raw copper lines at the exchange, while “bitstream access” refers to a number of wholesale PPPOE backhaul scenarios presented as an alternative to proper LLU.

In South Africa, ISPs wishing to provide ADSL services can either become pure reseller’s of Telkom SA’s end-to-end DSL ISP product, or they can purchase wholesale access to the ADSL “cloud” via a product called IP Connect. While it seems similar to bitstream access, Telkom has deliberately crippled IP Connect with a really ugly L3 routed hack involving three VRFs and a RADIUS realm per ISP, some RADIUS proxying and ISP-allocated address pools configured inside Telkom’s network.

Much has been written about the shortcomings of IP Connect, but the TL;DR version is that you need to tunnel to do anything other than the simplest best-effort IPv4 service that could work. What’s perhaps more interesting is to note that compared to the much more flexible bitstream service they could deliver using L2TP, IP Connect is very likely more expensive to run (in my opinion).

So, briefly back to regulatory matters: despite a government policy that LLU must happen, as is typical of our regulator, they have dragged their feet on implementing LLU regulation for many years. Telkom SA has cried wolf about LLU not being viable (as expected) and so the rest of the nation waits while we slip down the global competitiveness rankings.

With the recent replacement of the minister of communications, it seems there is some action (or noise?) around LLU (for example, see this article on Techcentral).

I predict that ICASA means business this time, but there is a catch: Telkom SA will offer bitstream access as a compromise to what they’ll present as the impending disaster of full LLU.

ICASA will fall for it, and Telkom SA will present a fully developed bitstream product as their “gift” to the industry (which was actually ready for launch in 2010, and will save Telkom money, compared to IP Connect).

What remains to be seen is:

1. Whether ICASA will regulate the price of the “units” of LLU? If I’m right, the “unit” will be bitstreams, not copper lines. In my opinion, the price must be regulated for bitstream access to be an acceptable substitute for full LLU (I’m not saying price regulation is a good thing – that’s a completely different question).
2. Just how far Telkom will push the interconnect point for ISPs away from the end user? The ideal scenario is access to the bitstream at the exchange. If they force interconnection with ISPs on a regional basis instead, they’ll make it less like full LLU (consequently defeat the purpose of opening up the access network to competitors), and artificially inflate the price by claiming the cost of backhaul to those regional interconnection points.

I hope I’m wrong and we get full LLU, but I have a feeling my prediction will be spot-on.