Hello Everyone, Adding to what I think Dorian just "hinted," traffic patterns in Asia are changing. If you are in the US, expect future where with equal or paid peering to places like China. In many cases, players in Asia are going to ask the peering partner to pay to cross the ocean to get to them. Remember, those companies who track the growth of the Internet with "English" surveys do not accurately count the grow in language contiguous regions. Some of my colleagues running networks in these "language contiguous regions" tell me of traffic patterns where 80% - 90% of their traffic is domestic while 10%-20% is "international." So like all peering exercises, who you know is the key to effective peering. IMHO - the best way to get the most efficient and cost effective peering with players in Asia is to go to Asia's equivalent to NANOG -> APRICOT. Attending APNIC meetings are also a extremely useful forum to network for peering contacts (in the past non-APNIC members just have to pay to cover their lunch). URLS: http://www.apricot.net http://www.apnic.net For now, use these guidelines: 1. Expect separate peering sessions for each of the Asia's core "language contiguous regions." English, Mandarin, Japanese, Korean, and Thai. Malaysian and Basa-Indonesian will pick up in the future -> but these can be covered with a mid-region peering partner (Indonesia is the #3 most populace country in the world). Each of the "language contiguous regions" have multiple L2 IXs or Transit Hubs (services like STIX). So you have options. 2. Expect to pay partial or total cost to get across sections of the ocean. 3. If you are looking for really good connectivity, expect a three zone coverage plus several US West Coast IXP drops: - North Asia: Covering Hong Kong-Philippines up North to Korea/Japan - Mid-Asia: Area surrounding Singapore, Brunei, Indonesia, Malaysia, and Thailand out to India. Because of the oceanic interconnects in Hong Kong, you can consider Hong Kong as a place that spans North and Mid-Asia. - Southern Asia: Down under with Australia and New Zealand. - US West Coast IXPs: A lot of players in Asia will not talk to you unless you have drops in one or more US West Coast IXP/Peering Centers. - Also start looking at a way to effectively reach Southern Asia. Mid-Asia points (Singapore Malaysia, Hong Kong, or Brunei) and UAE are key points for pulling peering from Southern Asia. Etisalat (UAE) is out to pull as much oceanic cable into the country as possible. So consider them the Singapore/Hong Kong/Japan equivalent west of Southern Asia. 4. Expect to pay to get to China. Tables are now reversed. There are two languages that span the entire planet and (from my observational evidence) impact Internet traffic patterns. English is one - which is why the US has been the focal point of the Internet. Chinese is the second - which is why players in China are asked why they should provide "free trans-oceanic access" at a US IX when +80% of their traffic is domestic (in this case domestic traffic = revenue). 5. Do not forget the big global IP players. Several of the big global players have their own backbones in Asia, sell capacity between Asian players, connect to at least one IXP in each of the countries, sell in those countries, and sell "Asian transit" to US/European players. 6. Do not be deceived by US West Coast IXs. Yes, there are a lot of Asian Players with boxes/links in those places. But that does not mean the Asian Players are going to peer for free. It is not 1995 any more. When you look at the top Internet per population figures, you will find places like Singapore and New Zealand in the top five. When you expand your scope to find indicators for Asia's Internet future, you will see where the region is going. Places like Hong Kong have more cell phones than POTs (I just read something where it was 2:1) with the cell phone penetration approaching a 1:1 figure for people age 18 - 60. Add these indicators to the fact that most of the Internet market dominating companies are the old PTTs. All these PTT (control freaks) are now Telcos (out to maximize share holder profit). All of them took a lot of hard knocks in the early years, learned from their mistakes, and now know where their markets are now going. So instead of dealing with clueless Asian PTTs in 1995 you are dealing with really clueful IP savvy Telcos in 2002. IP savvy Telcos who are members of the trans-oceanic cable businesses and are the ones buying up the excess capacity built by the failed "independent" trans-oceanic cable businesses. 7. Make sure you have up-to-date and accurate cable maps on your walls. When you get down to it, trans-oceanic Internet peering is all related to the trans-oceanic cable business. The "independent" trans-oceanic cable businesses have gone bust, but the "club" trans-oceanic cable businesses are thriving. My $.02, Barry
-----Original Message----- From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu]On Behalf Of Dorian Kim Sent: Thursday, September 26, 2002 11:43 AM To: keichii Cc: nanog@merit.edu; Neil J. McRae Subject: Re: AP IX locations
On Thu, Sep 26, 2002 at 01:30:42PM -0500, keichii wrote:
From: "Neil J. McRae" <neil@DOMINO.ORG>
I'm looking to improve my connectivity into the AP region, in a cost effective [i.e. for as little as possible :-)]. I have ruled out buying transit as it doesn't help the issue that I'm trying to resolve, so I was wondering if there was a location/IXP in the AP region that would enable me to interconnect with as many AP carriers as possible.
Above.net and AT&T are your best bets for operations based in
the Americas.
Above has a .jp IX/colo that is almost "the" best connected place in AP. ATT and Above.net provide almost 80 to 90% of the bandwidth from the Americas to AP.
A handful of incumbent telcos of AP region countries as well as few others operate multigigabit IP networks across the Pacific. I don't think they'd agree with your statement.
-dorian