Brian, Can this be done if the link to the ISP's are located at different Sites? This would be site to site load balance of outbound traffic using ISP #1 at Data Center #1 and ISP #2 at Data Center #2.
--- dre wrote: > ""Brian (273954)"" wrote in message > news:200210190024.AAA02320@;groupstudy.com... > > We are bringing in a second DS3 line into our > Cisco 7206 v12 router and > was > > hoping for some general advise. > > Our current provider is a 9mbps DS3 from Genuity. > We are bringing in a > > seconds DS3 from PAJO at 6mbps for redundancy and > to bring the usage down > on > > the Genuity line. > > > > My question is what commands do I need to look > into when I have the BGP4 > > setup on the router in order to handle the flow of > traffic properly? > > > > 90% of our traffic is OUTBOUND (up to the > internet) and we need to balance > > this traffic between the 9mbps and 6mbps > connections. From what I am told > > this won't be an automatic process but something I > will have to tweak on > > occation depending on if traffic demands change > between the connections. > > > > How am I best going to control this outbound flow > of traffic? My idea was > > that when one connection is using a high % I can > manually modify a metric > of > > some sort to make the router believe that that > connection is not as > > preferred as it once was and start sending a > little of the traffic over > the > > other connection instead. Am I being realistic > here? > > See RFC 3272 sections 7&8. In practice this > involves the following three > simple steps and there are only three ways to do it > (I'm trying to make this > easy): > > Ways to affect outbound traffic with BGP attributes: > 1) local-preference > 2) inbound as-path prepends > 3) inbound metric (MED) > > Steps to affect outbound traffic with BGP: > 1) Hard part (try to get your "Top N AS > destinations" using packet capture > + prefix lookup (ARIN, RADB) or NetFlow (easiest > method) with outbound > bits/sec first) - caution: there are many ways > to do this wrong or > incomplete, > see http://www.nanog.org/mtg-0110/tundra.html > for better ideas: > > Configure an "ip as-path access-list" using BGP > regular expressions and > destination ASes (e.g. _1_ or _701_ or _1239_ > ... etc) > 2) Easy part > > Configure a route-map to match your "ip as-path > access-list" and set one > of those BGP attributes above (read RFC 3272 and > understand the BGP > decision process before deciding what to use > where). You may want to > match on "ip community-lists" instead of "ip > as-path access-lists", but > you > need to create them first (see the Cisco IOS > documentation for further > details on bgp communities and community-lists). > > Note: the Internet routing table changes > constantly (you can see what > changes with "show ip route | include 00:00" > every minute or by looking > at your BGP updates -- requiring zebra bgpd or > Cisco debug mode). > Your route-map's view of the Internet and what > it matches also changes. > This requires somewhat constant updating (at > least you have to monitor > it). The best way of monitoring your circuit > levels is by using either > internal or external RMON polling (internal > means using the Cisco IOS > rmon command - setting up events and alarms to > threshold ; external > means collecting SNMP or RMON from the > interfaces and using an > external application, e.g. HPOV NNM, to set the > RMON events and > alarms). You will also note that your > route-maps may match differently > when a circuit resets or the router reboots. > This can be for a number > of > reasons, most of them beyond the scope of this > email. > 3) Careful part > > Configure the route-map on your neighbor and set > your inbound policy > (affecting outbound traffic) on your neighbor's > routes. > > E.g. router bgp x ; neighbor x.x.x.x route-map > AS1_OUTBOUND_TE in ; end > ; wr mem > clear ip bgp x.x.x.x (or ASN #) soft in > ( or just "clear ip bgp > x.x.x.x in") > > This is also detailed more in this new book: > http://www.oreilly.com/catalog/bgp/chapter/ch06.html > > But the author doesn't seem to understand some of > the logic in BGP-4 and > Inter-Domain > Traffic Engineering, especially in terms of > affecting inbound (I wouldn't > implement outbound > with local-preference all the time, and I wouldn't > ever announce > more-specifics without > reading RFC 2519 and "knowing what you are doing and > why you are doing it" > to affect > inbound traffic). But it's a well-written overview > of traffic engineering, > even though it > somewhat encourages worst-practice. > > Here are some really good resources on the topic: > > http://www.nanog.org/mtg-0206/te.html > http://www.nanog.org/mtg-0202/te.html > http://www.nanog.org/mtg-0206/feamster.html > http://www.nanog.org/mtg-9901/ppt/bgp102/index.htm > http://www.nanog.org/mtg-9811/ppt/avi/index.htm > http://www.nanog.org/mtg-0006/confed.html > > -dre [EMAIL PROTECTED] __________________________________________________ Do you Yahoo!? Y! Web Hosting - Let the expert host your web site http://webhosting.yahoo.com/ Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=55921&t=55918 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
