It’s very important to control inbound traffic when you are configuring BGP. icon_BGP.png If you have more than one ISP links, you should ascertain that your BGP announcement is peddled in the Internet. If you have only one link to the ISP, you don’t require BGP. Your requirements will be satisfied by a simple static route statement.
The main reason for you to have multiple links to ISP is reduction, failover and load-balancing traffic. Your BGP announcement is directly affecting your inbound traffic. BGP daily basis help you to escape real problems. The way of the routes announcing bring traffic on one of your uplinks.
Nowadays, ISP routing policy changes to avoid the loss of their customers traffic. ISP tags has high level of local preferences on their customers routes. It make traffic force to flow through their service link over the alternate path is learned from peer ISP. Morreover, Tier2 ISPs are tagging as higher local-pref on their customers route when they announce it to upstream providers(Tier1 ISPs) using BGP community .
Do not let ISP control your BGP announcement.
Do control of your inbound traffic.
Do not allow one of ISP to controll all your traffic.
Get involve more on traffic engineering.
Ask BGP community string to your ISP
Check your BGP announcement through public route server periodically
If you have two ISP links with owned /24 IP block that is allocated from InterNIC directly(but not from your ISP), it is no fun to control inbound traffic. B/C you cannot split it in half for redundancy plan.
See below example,
Let’s assume your public IP address is 172.16.1.0/24 and two ISP link. If you announce route with below plan, it would cover redundancy(failover) and load-sharing
ISP “A”
Announcing 172.16.1.0/24 and 172.16.1.0/25
ISP”B”
Announcing 172.16.1.0/24 and 172.16.1.128/25
ISP “A” will take most of traffic that is destined to 172.16.1.0/25 and guaranteed within ISP “A”, but not outside of ISP “A”. ISP “B” will take most of traffic that is destined to 172.16.1.128/25 and guaranteed within ISP “B”, but not outside of ISP “B”. This is just example. There are quite a bit ways to control inbound traffic.
Look around our BGP sample configurations and article for discussing inbound traffic control
The main reason for you to have multiple links to ISP is reduction, failover and load-balancing traffic. Your BGP announcement is directly affecting your inbound traffic. BGP daily basis help you to escape real problems. The way of the routes announcing bring traffic on one of your uplinks.
ISP shouldn’t control your traffic
Why one link takes more traffic than others? It’s very interesting question. Moreover, he main reason is in the ISP routing policy changes. WHAT~ How could they.Nowadays, ISP routing policy changes to avoid the loss of their customers traffic. ISP tags has high level of local preferences on their customers routes. It make traffic force to flow through their service link over the alternate path is learned from peer ISP. Morreover, Tier2 ISPs are tagging as higher local-pref on their customers route when they announce it to upstream providers(Tier1 ISPs) using BGP community .
Do not let ISP control your BGP announcement.
Do control of your inbound traffic.
Do not allow one of ISP to controll all your traffic.
Get involve more on traffic engineering.
Ask BGP community string to your ISP
Check your BGP announcement through public route server periodically
ISPs are not accepting smaller than /24, c class IP block
To reduce BGP routing table size, most of the ISP is not advertising smaller than /24 IP block to peers. In other words, if you have your own IP address /24, and split into two /25s. Announcing first half of /24 to ISP A and second half of /24 to ISP B, then you will have reachability issue. Because, ISP A wont announce first half of /24 to ISP “B” vise versa. You must announce bigger than /24 IP block if its your own.If you have two ISP links with owned /24 IP block that is allocated from InterNIC directly(but not from your ISP), it is no fun to control inbound traffic. B/C you cannot split it in half for redundancy plan.
See below example,
Let’s assume your public IP address is 172.16.1.0/24 and two ISP link. If you announce route with below plan, it would cover redundancy(failover) and load-sharing
ISP “A”
Announcing 172.16.1.0/24 and 172.16.1.0/25
ISP”B”
Announcing 172.16.1.0/24 and 172.16.1.128/25
ISP “A” will take most of traffic that is destined to 172.16.1.0/25 and guaranteed within ISP “A”, but not outside of ISP “A”. ISP “B” will take most of traffic that is destined to 172.16.1.128/25 and guaranteed within ISP “B”, but not outside of ISP “B”. This is just example. There are quite a bit ways to control inbound traffic.
Look around our BGP sample configurations and article for discussing inbound traffic control
