Border Gateway Protocol (BGP) is the map of the Internet.
Border Gateway Protocol Definition
Border Gateway Protocol (BGP) is the protocol of the Internet. It is for exchanging routing information in a network of autonomous systems (AS). The autonomous systems are gateway hosts, each with their own router. BGP contains a routing table with a list of routers and addresses it can reach. Along with addresses, it contains a cost metric for each router. This allows BGP to select the shortest logical path by default.
The map, as created by BGP, then defines to router A which part of the Internet is reachable over router B. Router B might take care of delivering the data inside of its own network. Or, pass it on to another router, C, at another ISP that knows how to reach the destination.
BGP is a complex network of protocols. This determines how traffic moves through the Internet. This is also the protocol used to communicate between different ISP’s. It allows global Internet transit.
BGP is a complex system, which is designed for scalability. You can see this as the BGP system has grown over three times since 2003. With this growth, the capacity of BGP routers has been able to handle the extra memory.
Meaning, the system has the ability to scale.
The Internet’s Default Route
BGP sends around 95% of traffic over best path selected routes. These routes are based on the most logical route. This means traffic is sent over the shortest logical route. This does not always mean the best possible route (taking into account latency and bandwidth), nor does it mean the most cost effective route.
Best path selected BGP routes can become congested. They also may not have the best bandwidth for specific data transfers.
This issue becomes the problem with relying on the default BGP routes. Particularly when needing low latency. A system will need to override the default BGP route to provide the best possible route to the end user.
Override Best Path BGP
Network optimization has three areas of consideration: latency, bandwidth, and cost. These three considerations are not accounted for by the default BGP best path selection. BGP cannot handle this as it is not an intelligent routing system. It has no mechanism to send traffic over different routes.
When considering latency, the most logical route for traffic may not be the lowest latency route. Routes may have congestion, or lack enough bandwidth for the route to be the fastest.
Bandwidth becomes the second item of consideration. This has a lot to do with the amount of traffic over a given route. Also, route capacity is a consideration. Reason is there could already be significant traffic on the route.
Lastly, optimizing for cost is another important consideration. When speed and latency aren’t as big of a concern, reducing network operation costs can have its advantages. As Internet transit costs can be determined from border gateway protocol, routes can be optimized for this metric.
BGP is an important concept when looking at how to optimize network traffic. This brief understanding of what BGP is, and how it functions, can provide insight in how to determine the best options for Internet traffic.
To learn more about latency optimization, you can download our DevOps Network Guide Ebook.