However, despite all the relevant redundancy technologies that you can put in place, there is always concerns about controller failures etc. Looking at what Pluribus Networks is doing, they seem to address some critical areas of the future network design with the following.
1. Clustering - Building smart and open clusters allow easier management and co-ordination of network resources within the data center.
2. Network Hypervisor - The development of the Switch OS as a basic hypervisor creates a lot of opportunities to do many more things with NFV and other applications.
You can read more about Pluribus Networks.