Network Virtualization & Visibility

Many focus on the lack of visibility in network virtualization environments. It's time for more concrete conversations in this area.   The statements around visibility are usually too broad After a quick search, I found that Riverbed’s Cascade network performance management (NPM) solution already supports VXLAN and I’m sure what they offer will only get better.  That means they can let you know what applications are being used within the overlay tunnels.  A demo of the solution is below.  

Most modern vswitches also support some type of NetFlow/sFlow – are you leveraging that functionality today or just planning on it in the future?  Have APIs been examined to see what information can be extracted from the controller in the newer solutions?  When deeper visibility is needed today in a pure physical network, do you look to the network devices on a hop by hop basis to get data or at NPM/APM/NMS tools for triage and correlation?  Maybe as an industry, we can put pressure on companies like Riverbed, Solarwinds, Compuware, Netscout, etc. to help bridge this gap.

Here is another link worth taking a look at about Riverbed monitoring SDN/VXLAN environments.  http://www.slideshare.net/riverbedtechnology/riverbed-cascade-and-vxlan-monitoring

And for those large organizations that have private MPLS backbones, are there LAN and WAN teams?  Can the LAN team see every hop taken within a trace route between two Campus sites?  Why was CAPWAP with WLAN never an issue especially with Layer 3 roaming? 

Is the bigger challenge about network vs. server admin and point of control/power?

That all said, I really do want to collect data on where people think visibility needs to be better in the world of network virtualization.  Please write in, comment, or email.

Thanks,

Jason (@jedelman8)