Rising to the Top: A Baker’s Dozen

As readers of this blog will know, Renesys collects Internet routing data — a lot of it. We use this data in a variety of ways: in determining the impact of cable breaks, natural disasters and deliberate partitionings; in uncovering the source of hijacks or other questionable activity; in analyzing Internet business relationships; and in exploring “what-if” scenarios.

All of our reports and products are based on hard facts and objective analysis. Perhaps the only controversial thing we do with our data is to rank all the service providers in the world: globally, by geography, and by market segment. The rankings are a rather crude measure of size, as they are based entirely on the quantity of IP space ultimately transited by each provider. Although there are obvious shortcomings in this approach, it is certainly objective and the process is fully automated. It also happens to be derived from data that is readily available for all providers. Routing data, unlike most other metrics we could consider using, is inherently public.

While everyone wants to be #1 (hence the controversy around rankings), changes in rank can be far more revealing than the actual rank itself. In other words, while there are surely big differences between #1 and #50 in our AS rankings, the differences between #5 and #6 are much less clear given the nature of the metric. What we tend to look for are abrupt changes and long-term trends. Did a provider just jump in the rankings? Maybe they picked up a large customer or a nearby rival lost one? Who was it? Is another provider showing steady gains in the rankings? Maybe they are consistently taking market share with an aggressive, well-executed business plan in a particular part of the world? This is why changes in rankings matter: they capture some of the dynamics of the business of providing Internet service. With this in mind, we will take a look at the top 13 providers in the world for 2008 and how they have jockeyed for position throughout the year. We will also highlight some of the more interesting changes.

To rank providers, Renesys starts out by determining the business relationships between adjacent Autonomous Systems (ASes), as seen via BGP announcements from our over 300 peers. This is a complex, computation-intensive task, consisting of processing over 10 million AS paths and 100,000 unique AS pairs that make up these paths. By combining the computed business relationships with IP prefix geo location information, we can determine exactly which prefixes are ultimately transited by a given provider in a given location. Rankings are then based on the scores we assign to the relevant prefixes, scores which depend on prefix size and date of assignment. (More details can be found here.) In this blog, we’ll concentrate on the top 13 providers according to our global rankings and how their score changed over 2008. Since the exact numeric score is not that meaningful in this context, we will not provide the scale in our graphs. Rather, we will show how the scores changed over the year and consider some of the reasons behind these changes.




The above graph plots global scores over time for the top 13 providers. While this might not look very informative, if you squint carefully enough, you can see three distinct clusters of providers, namely, the top two, the middle six, and the final five, each operating within roughly the same range of scores over time. We will consider each cluster in turn, enlarging the scale so we can more clearly see the changes, and then making note of some of the more interesting ones.



In 2006, when Renesys first started ranking providers, Sprint (AS 1239) seemed to have the #1 spot locked up with a score of almost 25% more than second-place Level 3 (AS 3356). Since that time, Level 3 has been gradually and consistently closing the gap and took over the #1 spot for the first time this year. Despite a few long, relatively flat periods, Level 3 had two huge surges, one in July and another in November. In July, Level 3 picked up Korea Telecom (AS 4766) and ReTN.net (AS 9002) as customers. (Previously we had them classified as peers.) And Asia Netcom (AS 10026) greatly increased the number of prefixes they were transiting via Level 3. November’s rise was almost entirely due to Japan’s Softbank Telecom (AS 4725) increasing the number of transited prefixes from 2% to just over 11%. Although they seem to have tailed off recently, Sprint’s growth was more consistent over the year, but not enough to counter the big wins by Level 3. The average score in this cluster increased by 14% over the year.



Looking at the second cluster of providers, a few things are readily apparent. First, Verizon (AS 701) and NTT (AS 2914) both suffered sharp declines over short periods. Second, Global Crossing (AS 3549) and TeliaSonera (AS 1299) grew consistently during the year and were the only ones in their peer group to show appreciable gains. Verizon lost big after TeliaSonera stopped using them for transit. (Peers, whether paid or not, do not get credit for each others customers.) Likewise, Cogent became transit-free mid-year after they stopped using NTT to reach AOL ATDN (AS 1668), which accounted for much of NTT’s decline. Global Crossing’s gains consisted of numerous big wins, many in Asia, picking up increased transit from Hurricane Electric (AS 6939), PCCW (AS 3491), ReTN.net (AS 9002), Asia Netcom (10026) and Golden Telecom (AS 3216) to name a few. As a result, Global Crossing catapulted from #5 to #3 in the world, quite an accomplishment, especially considering they were #25 back in 2006. TeliaSonera also picked up increased transit from Hurricane Electric and Asia Netcom, briefly passing AT&T to capture the #7 spot for a time. But as of a few days ago, AT&T sneaked ahead again by picking up more transit from Uninet (AS 8151), a large retail provider in Mexico. Still, given the flat or declining fortunes of most of this group, I would not be surprised to see TeliaSonera ranked #4 or #5 by the end of 2009. While quite small, the average score increase for this cluster is not very meaningful, given the very different trends observed.



The members of our final cluster largely tracked one other with similar growth rates. One exception is Tiscali (AS3257), which moved from the bottom of its peer group to near the top, within striking range of Tata/VSNL/Teleglobe (AS 6543). In fact, they surpassed them briefly during the year on a couple of occasions. Tiscali picked up increased transit from Asia Netcom and gained Interoute (AS 8928) and KPN (AS 286) as customers. China Telecom (AS 4134) moved to #11 all the way from #21 back in 2006. While China Telecom is by no means a global provider, they are by far the dominant player in the huge Chinese market, and also provide transit to South Korea and Vietnam. These large and emerging markets allowed China Telecom to move up two places in our global rankings this year and to within easy striking range of the top 10. The average score in this cluster increased by over 40%.

Conclusions

It’s pretty clear from this narrative that the providers who are moving up in the rankings are doing so thanks in large measure to Asia. This should come as no surprise. With more than half the world’s population and only 15% online, there is enormous potential here. But market share does not imply profitability or even a well-run network. So we would never suggest you pick your providers based on Renesys Market share rankings alone. While declining market share might indicate a problem, a growing company is worth your business only if they meet your service needs and are going to be around tomorrow. Rankings, by their very nature, cannot take all possible considerations into account, nor can they be tailored for each consumer’s needs, and thus they should only be one factor used in making an informed business decision.

5 comments
Dwayne
Dwayne

Where is Sprint picking up its growth? Asia also? Are they getting it elsewhere? Editor's Note: Sprint's score can increase by having its existing customers grow or by picking up new customers. An example of the former is Tiscali (AS 3257), which transited many more prefixes via Sprint in 2008. As an example of the latter, US DoD (AS 721) picked up Sprint as a new provider in May. For this blog, we did not preform an exhaustive analysis of each carrier, but instead focused on a few major changes during the year.

Pat
Pat

This is good stuff for a view of macro network usage growth demographics. Clear to the blind are the pointers to network access capable consumer products in Asia as the market for sales and market share emphasis. I love the obvious! Especially when stated so clearly.

Dan
Dan

Why is Verizon only considered AS 701 and not the sum of all of their ASNs (and transited prefixes), considering that they (the old UUNet) us a multi-AS structure, e.g., AS 702 in Europe? Or does the graph cover all of UUnet but just refer to it simply as AS 701? Editor's note: This is a good and fair question. 701 is a bit of a special case in that they use different ASes in different parts of the world. Of course, other networks use multiple ASes, but to a much lesser extent. To keep things simple, we used the largest AS and all prefixes transited through it. For reference, on Dec 19, the score of 702 is only 15% of 701, while 703 is 7% of 701. While significant, these additions do not overtake 3549 in the rankings. The blog is focused on major changes, not so much on the exact ranking.

Brandon
Brandon

Just a layman here, but there's one conclusion I draw from the first graph which wasn't mentioned: TeliaSonera moved from the 3rd cluster to the 2nd cluster, over the course of the year. Editor's note: Excellent point. You could also say that Verizon left a cluster of its own and entered the 2nd cluster from above.

buzzup.com
buzzup.com

Rising to the Top: A Baker's Dozen As readers of this blog will know, Renesys collects Internet routing data — a lot of it. We use this data in a variety of ways: