Undersea cables are expensive to install. But if you're an Asian Internet hub
trying to connect to other Asian Internet hubs across un-cabled waters, what
else can you do?
Well, one alternative we see is Internet Providers heading to California, as many Asian providers opt for
Internet paths out of Asia to the west coast of the US, and then back to Asia.
These tortuous routes, aptly called
hair-pinning
(observe their supple shapes), may be cost-effective initially, but generate
latency, which can be a problem for some businesses (and their end users).
Well, one alternative we see is Internet Providers heading to California, as many Asian providers opt for
Internet paths out of Asia to the west coast of the US, and then back to Asia.
These tortuous routes, aptly called
hair-pinning
(observe their supple shapes), may be cost-effective initially, but generate
latency, which can be a problem for some businesses (and their end users).
For more than a year, Renesys has been collecting daily latency data from 30+ locations to every prefix on the Internet. As you can imagine, we've learned a lot.
For example, here's how Australian carrier Vocus routes from Sydney to California and then back through Asia to reach India:
Who's Number One?
Such nimble routing schemes can produce surprising effects on our Market Intelligence analysis of Asia. Some US-based carriers with little infrastructure in Asia have higher Renesys Wholesale and Backbone Rankings because they have lots of Asian networks interconnecting via the USA. Level 3, for example, has a modest Asian infrastructure, but due to its Asia/US network connections, topped our Asian rankings even before they acquired Global Crossing.
Asian carriers are beefing up their intra-Asia networks to satisfy
increasing demand for local Asian content (and a corresponding decrease in
demand for US content.) Renesys is developing a model to determine the most
influential NSPs in this intra-Asian market. However, since BGP routing
tables don't tag routes by geography, it can be tricky distinguishing between
wholesale business based on Asian customers within Asia, and Asian customers
connecting via the US. So we adapted our Asian Retail Rankings algorithms to
differentiate US and Asian wholesale NSPs. TelecomAsia
recently published
our initial rankings as the first quarterly insight into the intra-Asia
market.
Jockeying for Position
Several distinctive characteristics excluded PCCW, one of the largest Asian networks, from our initial intra-Asian rankings. PCCW's success in Asia comes mostly from their backbone business rather than retail and wholesale endeavors. Their network is nearly settlement free (6% of their weighted routes go to their provider Global Crossing, while Pacnet routes 57% to their providers), which made it resemble large US Tier-1 providers - and why it didn't appear in our previous rankings. Rapid growth should catapult PCCW into Tier-1 status soon.
PCCW's backbone bias got us thinking about how Renesys evaluates Asian
providers. We analyze three rankings to calculate total customer size:
Retail, Wholesale and Backbone Rankings are correlated to
approximate the corresponding business models of IP transit providers.
Depending on which ranking we examine, very different pictures of Asia emerge. For example, Chinese and Japanese NSPs with retail business divisions dominate Retail Rankings because their huge market size influences their Total Customer Ranking. US providers dominate the Backbone Ranking because they generally connect large regional or national incumbent providers' wholesale customers to each other and also to US content.
It's enlightening to throw a few extraordinary cases into the mix. China Telecom is by far the largest retail provider in the world (not just in Asia), but is ranked 4th in Asian wholesale and 26th (including non-Asian NSPs) in Asian backbone. NTT's international backbone network is ranked 24th in the Asian Retail index, 1st in the Wholesale Index and 4th in the Backbone Index (including non-Asian NSPs). When you combine these three rankings, NTT (AS 2914) is about 20% bigger than China Telecom (AS 4134)!
Here's the distribution of a representative sampling of providers based on their Wholesale to Retail bias:
While not a quantitative comparison, by clustering providers according to their relative rank in the Retail and Wholesale markets, we can group them into functional categories. Providers in the green segment are strong in both retail and wholesale rankings. Yellow-segment providers are strong in wholesale (typically in multiple countries). Providers in the salmon-colored segment are strongest in backbone and wholesale.

Hi Bob,
Nice article. FYI we have just installed a POP in Singapore and will be going live in the next 14 days. You should start to see AS4826/Vocus routing traffic on the direct Perth-Singapore path shortly after!
McDonald Richards
CTO
Vocus Communications
AS4826
Hi Bob,
Thanks for your clarification on the Blog.
The blog could explain the complicated nature of Asia ranking, and present a more accurate and reliable view.
On the Asia ranking I suggest Renesys to include a parameter to measure how good a transit provider can maintain Asia traffic to stay within Asia. This is becoming more important parameter than just counting the number of routes (example, I doubt the top AS in your revised list, their traffic with CT/CU is still predominantly route via US)
Understand there is challenge as the BGP route does not indicate the physical path. But you mention that Renesys does perform measurement of actual destinations latency. Not sure whether this could be share and build into the ranking.
Louis Ho
AVP IP Products
PCCW Global Ltd
AS3491
Louis,
We are looking for ways to help people with our extensive latency data. The first product will likely be an overview of the Internet market to market long-term latency www.renesys.com/tech/datasheets/IPHP_Datasheet.pdf
Bob Fletcher