Earlier this summer Apstra announced AOS - a vendor agnostic datacenter automation platform designed to simplify the complexity of service creation in the datacenter as well as the broader SDWAN.  The company offers an interesting and innovative solution to a problem that has, to date, hindered the growth of software defined WAN services.

This blog isn't an endorsement for Apstra, although we do think it cool, but it does serve as an example of another firm recognizing the problems with today's software centric way of looking at the network. 

Datacenter engineers are typically Linux savvy and understand the administration, care and feeding of large complex Linux environments.  Network Engineering - heretofore SDWAN engineers - understand the complexity and nuance of wide area networking.  Typically router centric, traffic centric, engineering focused on the WAN itself and not anyone physical location.

Its like two ships passing in the night - Linux savvy administrators are not versed at all in the way of WAN protocols or wide area networking.  Conversely the router jockeys running the WAN network are not at all versed in the way of Linux in general, much less the finer aspects of scripting in Python.  Together the groups have the skill set to run end-to-end services.

We at SDNNFV.NET believe this is a big reason why OpenStack, so successful with compute and storage, has not been as efficient with its success in the network component.  It requires a different engineering discipline than does compute and storage.

Hopefully we can share our collective knowledge, WAN engineers learning Python and Linux engineers learning there is life after IPTABLES.  Until then its "Who's on first?"