You Won’t Get Better Internet Until Old One Is Broken. 500K BGP Routes Good Start.

Some people have pointed out the Internet BGP table is now at 500,000 IP Routes. If you people don’t hurry up and blow this to a million entries, we will never get decent routers and greater bandwidth in the carrier backbones.

BGP Routing Table - 500000 routes

CIDR Report – AS2.0 BGP Table Statistics

At 500K entries, many of the decade old routers (like the Cisco 7200) must now be replaced. That’s a good start.

This is a good thing for customers since newer routers go faster, have more features and more reliability. It will drive those lazy service providers to invest in processes to rotate their assets more often and reduce the price of updates and bandwidth.

So get out there and advertise all of your /20 as /24′s and burn up the global routing table. Once we blow that sucker up, people will also starting taking IPv6 more seriously and quickly add hundreds of thousands more entries.

More is just more when it comes to bandwidth and the Internet.

Bandwidth solves all networking problems - EtherealMind.

  • Darin

    “Too many BGP entries would never be enough” ?

    • Name Admin

      Fodder for the mind…albeit probably sarcastic… IPv6 will solve a plethora of issues and bring better QoS into the mix…

  • http://www.leafpoint.net danmassa7

    Having all our gear contain all routes to all possible locations on the entire planet seems like a really bad idea to me. Why does an ISP need 500k routes? It doesn’t scale.

    We need a protocol that aggregates prefixes into Forwarding Equivalence Classes (FECs) or some other layer of abstraction.

    So, if you have twenty connections to other autonomous systems where you are peering or buy transit, and you have 90,000 customers, you only have 90,020 FECs. A Catalyst 6500 with the original Sup720 can hold that in its ASICs, no problem.

    Maybe when the IETF starts working on IPv7 they will make something like LISP a mandatory requirement! We will never run out of IPv6 addresses… because the IPv6 routing paradigm will reach its limit much quicker.