Editing DNS API (section)

== Why is my traffic ending up in another node? ==

Our Anycast networks use BGP to distribute traffic to different upstream providers around the world. These upstream providers range from Tier 1 level, the likes of Cogent, NTT, GTT, or others, and lower tiers which are local in a continent or a country, and traffic via Internet Exchanges which is either by using their route servers or through direct peering. 

A BGP relationship is best effort by default, and once traffic leaves the premise it will be in the *total* control of upstream providers. Some upstream providers have pre-defined agreements that will influence the route the traffic will take once our packet leaves our network. 

Due to loss of control of traffic once passed on to upstream providers, DNS clients traffic might be forwarded to nodes that are quite far in both terms of network connectivity and geographical distance.

For instance, a client in North America could see their DNS query routing towards a node in South Africa, when in fact the query should have been directed towards the many nodes available in North America.

Many providers will change their internal routing configuration after we contact them if we find an issue, but big players in the market like Hurricane Electric, never do so. Whilst we stopped our direct peering with them, many of our upstream providers use them as one of their own upstream connections.

Our DNS team continuously tracks those issues and contact suppliers where possible. Most of the time those issues get resolved, however it always takes time. We also employ the use of BGP communities where applicable, to influence how our upstreams route traffic.