It is common wisdom that BGP is awesome because, you know, … something. It runs the Internet therefore it must good. It can be extended and that makes it good. Lots of vendors use it …..
Hang on. When you think about it there are no good, solid BGP implementation in open source. Quagga is well known but I’ve not heard anyone claim that it’s mission critical. Juniper and Cisco are always quick to claim that BGP is the most robust protocol. Juniper loves BGP so much that it its Contrail product is basically SDN over BGP. (alert! – gross oversimplification)
In conventional logic, there is not much value in owning a BGP protocol stack. Open source software is usually most successful when the functionally is a commodity function like the Linux operating system, OpenDaylight controller or OpenStack cloud orchestration. Having dozens of competing commercial products doesn’t always grow the market, more often it stifles and prevents overall market growth.
So why are there no world-class, proven and reliable open source implementations of BGP ?