Hi guys,
I'm still digging to give corrects answer to my manager who is puting pros and cons for nutanix on the table.
As I need to make a design without knowing exactly where he want to go, I'm making assomption on it. I understood that they wanted to put NSX on it VRA and vCloud SP.
I've found a couple of recommandations / tips on NSX over NUTANIX but nothing that can be used on design part (more on implementation which is quite cool !) Do someone have integrated it on a 4.5 cluster ?
do Nutanix plan to give us a technical note / any other cool stuff on NSX ?
sheers,
Solved
NUTANIX & NSX

Best answer by vcdxnz001
Hi ,
I'm doing the testing for NSX with Nutanix that will be documented in an upcoming tech note. I'm also planning to put a blog together once i've done the testing. The results so far are very good. At a high level Nutanix is invisible to NSX and it really just works. The recommended versions are NSX 6.2 with vSphere / ESXi 6.0 U1a (assumes NSX-V). This gives the best features and also the highest performance. I have tested 4.8GB/s, i.e. line rate, per host with 2 x 10GbE NIC's (scales linearly) and latency of 76us on average between nodes.
The design patterns for NSX in terms of the vSphere clusters will be the same as VMware recommends, Management Cluster, Edge Services Cluster(s), Resource/Compute Cluster(s). Installation / Configuration is also the same. Also as our CVM is just a VM, it also works on an NSX vWire.
Important considerations are MTU size on the underlay network. In my environment I'm using 9216, and I'm using 9000 on the vDS. Due to the VXLAN overhead this makes the maximum MTU of any VM = 8950 bytes. In my case the CVM's are configured with MTU 8950 and are on a NSX vWire/VXLAN stretched between two different ToR switches. I've configured a leaf spine network architecture with L2 ToR to L3 Spine.
In terms of platform model selection the 1065G4 is a great choice for management clusters and ESG clusters. 3060G4 and 8035G4 or any of the other models make great resource/compute clusters.
Nutanix works with vRA, vCloud SP etc, so nothing changes there, again the platform is invisible to them. The value that Nutanix brings is a greatly simplified architecture that is quick to deploy, expand on demand, very simple to manage, and self heals when things go wrong. Deploy in minutes, one click upgrade at lunch time, spend more time doing what you want to do.
If you have any specific questions I'd be happy to answer them.
View originalI'm doing the testing for NSX with Nutanix that will be documented in an upcoming tech note. I'm also planning to put a blog together once i've done the testing. The results so far are very good. At a high level Nutanix is invisible to NSX and it really just works. The recommended versions are NSX 6.2 with vSphere / ESXi 6.0 U1a (assumes NSX-V). This gives the best features and also the highest performance. I have tested 4.8GB/s, i.e. line rate, per host with 2 x 10GbE NIC's (scales linearly) and latency of 76us on average between nodes.
The design patterns for NSX in terms of the vSphere clusters will be the same as VMware recommends, Management Cluster, Edge Services Cluster(s), Resource/Compute Cluster(s). Installation / Configuration is also the same. Also as our CVM is just a VM, it also works on an NSX vWire.
Important considerations are MTU size on the underlay network. In my environment I'm using 9216, and I'm using 9000 on the vDS. Due to the VXLAN overhead this makes the maximum MTU of any VM = 8950 bytes. In my case the CVM's are configured with MTU 8950 and are on a NSX vWire/VXLAN stretched between two different ToR switches. I've configured a leaf spine network architecture with L2 ToR to L3 Spine.
In terms of platform model selection the 1065G4 is a great choice for management clusters and ESG clusters. 3060G4 and 8035G4 or any of the other models make great resource/compute clusters.
Nutanix works with vRA, vCloud SP etc, so nothing changes there, again the platform is invisible to them. The value that Nutanix brings is a greatly simplified architecture that is quick to deploy, expand on demand, very simple to manage, and self heals when things go wrong. Deploy in minutes, one click upgrade at lunch time, spend more time doing what you want to do.
If you have any specific questions I'd be happy to answer them.
This topic has been closed for comments
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.