Inline On Thu, Dec 16, 2010 at 4:00 PM, Charles N Wyble <char...@knownelement.com> wrote: > It's not about access networks peering. That's usually not worth the > effort for the reasons you outlined below. It's about peering with the > content provider networks. >
All peering is good peering (until egos get involved). Peering with content providers may save you some money. Open peering with anyone sufficiently clued to have a ASN makes the internet and your provider community stronger. >> >> The current impediments to small ISPs peering are: >> 1. BGP skills and hardware. It used to be the only reliable thing for >> BGP was a big cisco decked out with overpriced ram. Now anyone can do >> BGP private peering with a PC running MT/vyatta/linux or an >> MT routerboard, or their cisco or their juniper. Still, few have BGP >> experience to do this comfortably. > > The level of effort is hopefully nothing more the a textbook templatized > config that connects you to the fabric. The talent is in running the > fabric. > Agreed on a textbook template. Disagree that any talent is needed to run the fabric. Setup some basic port security on a L2 switch (one mac address, etc) and get rolling. You don't need route reflectors or anything fancy to get started. Actually, you don't even need a shared switch if there are only two participants. >> >> You can get the talent in socal, but it's not nationwide. People could >> hire Butch or someone on guru.com to setup bgp, but they like to have >> the self sufficiency to DIY in many cases. I've probably met face to >> face all the people in my state who are proven BGP skillful and it's not >> a lot. > > Yeah it's a small subset for sure. In my experience if someone doesn't have the clue there are other ISP peeps in the area with clue and care to help. Perhaps our local small IX just has a good community. >> >> 3. decreasing uplink costs. Used to be you'd do anything to save a >> precious megabit and peering was one such thing. I had a satellite >> receiver system for receive usenet to offload the bandwidth back in >> 97ish. Now it's just outsourced. We used to cache a lot more web traffic >> too. Now it's helpful but not so important. If there were an occasional >> megabit of traffic going to another local ISP, I wouldn't really >> consider it worth the effort of peering. I would suspect most of the >> traffic between WISPs is email and a little random p2p, and perhaps some >> vpn activity between employees and businesses that use different service >> providers. The peers despite the extreme minimalist financial investment >> should be more reliable than the uplink to make good sense as well. > > Again it's not about access networks. It's about content networks and > access networks. We run ~5 Mbps on average with peaks over 50 Mbps.This is on a IX with 15 participants, most of whom you've never heard of. We have people working from home with a VPN to work. Online backup with servers that consultants host with other local providers (and servers with us and clients on other providers), VoIP systems, video conferences with teaches from schools. All of these applications benefit from local peering. See: http://kb.wisc.edu/ns/page.php?id=6636 and http://stats.net.wisc.edu/madix.html Please, everyone, consider the locations you have in common with other providers. If you are in a datecenter or larger facility check out http://www.peeringdb.com Peering will help more than it can hurt. -------------------------------------------------------------------------------- WISPA Wants You! Join today! http://signup.wispa.org/ -------------------------------------------------------------------------------- WISPA Wireless List: wireless@wispa.org Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/
