So I've been struggling to create a /16 network range as part of an External Network profile within vRA. Not getting any errors within the catalina.out log file on the vRA appliance and no errors within the IaaS Windows server event logs. Just for fun I decided to do some tinkering.
I have a NSX Edge Service Gateway already deployed with both the external and internal interfaces configured appropriately (Ext is /24 and Int is /16) and L2 switch deployed and connected to the Int of the ESG.
1. I created an External Network profile with no network ranges attached - Worked just fine
2. I edited the network range to accommodate a /16 network range, 172.16.0.1 - 172.16.255.253 w/ a gateway of 172.16.255.254 (Int interface). - No dice. Received "Service Temporarily Unavailable (503) error. It also took about 3-5 minutes to display the network range on the main page where I can click OK to commit the changes and save the network profile.
3. Deleted the External Network profile
4. Reconfigured ESG Int interface for a /24 subnet (172.16.0.254/24).
5. Deployed a new External Network Profile for the /24 subnet, a gateway of 172.16.0.254, and an available IP range of 172.16.0.1 - 172.16.0.253 - WORKS!!!!!
My vRA appliance has 2 procs, 8 GB of RAM and thick provisioned disks (saw a post about poor disk IOPS on thin disks causing I/O errors and the 503 error). This appliance runs at no more than 5% CPU and 40% RAM.
My IaaS server is 2012 R2 with 2 CPUs and 4GB RAM with thin disks. This server runs at no more than 50% CPU and 15-20% RAM.
Why I want to create a /16 network profile: Since I'm doing all this in a lab I don't need a "transport network" that supports up to ~16.77 million IP's, but I figured if the option was there, why not take advantage of it.
What's the largest transport network you all have created? What bottlenecks have you seen that you need to adjust for in order to provision networks of that size?
Jeremy