Benefits of Nexus 1000v

[TGAMBON_2]

In my UCS Bladecenter environment, we thinking about changing away from our standard (non-distributed) vSwitches to Nexus 1000v distributed switches.  We've been told that doing that is going to give us a clearer view of the network traffic paths from between virtual guests and their sources/destinations but looking at the docs and forum, I haven't really seen any information to support that so far.  Most of what I'm seeing indicates that there are advanced features of phyiscal Cisco switches will be available once you are using 1000v, but we're not really using advanced features (like QoS, Traffic shaping, PVLANs, etc...) so I'm not really sure it is going to benefit us.  The clear reasons I could see to move towards distributed switching are for ease of VLAN configuration per host and if you were in danger of hitting any of the limitations associated with standard vSwitches.  I'd appreciate any feedack to help guide is in making the decision on whether to change.

[Robert Burns]

Some additional features you might find useful are included in a post I did here:

http://communities.vmware.com/thread/316542?tstart=0

One of the key benefits of the 1000v, is allowing your VMs to be managed identically to how your physical hosts are managed.  Same CLI interface most network admins know & love.

Any DVS (Vmware or Cisco) is a improvement over the standard vSwitch.  Cisco just offers a great deal of more features & experience in switching than VMware's vDS provides. 

There are some great video's available on the 1000v overview & features here:

https://communities.cisco.com/community/technology/datacenter/nexus1000v?view=video

Let me know if you have any other questions after checking them out!

Regards,

Robert

[Walter Dey]

I agree 100% ! however I see many customers (and some VMware techies actually recommend this), who use classical old fashioned vswitch and DVS; vswitch for management interface, maybe vmotion, and DVS for user VM traffic.

I have seen DVS failing, which can end up in a disaster, if you don't have a proper backup.

[Jason Masker]

If it is a "clear view of . . . network traffic paths" you're after, then 1000v is the way to get just that. Netflow and SNMP capabilities are terrific on the 1000v platform. For a long time now with VMware, the closest view you had to individual server traffic was in aggregate at the host port. You will have visibility into traffic flowing to each individual server's virtual port. You also get 64 ERSPAN sessions per VSM which will give you incredible visibility for troubleshooting within the VMware environment

Cisco emphasises advanced features like QoS & PVLANs because these features are essential, in many ways, to cloud providers seeking maximize gains from virtualization.

Distributed switches alone do a lot for ease of configuration and scalability in the VMware environment if you need only basic switching similar, but I would argue that the visibility into individual virtual server network activity as well as the span capability and tools within the 1000v at your disposal for troubleshooting more than justify the delta in cost for many enterprise environments.

[TGAMBON_2]

Great information, guys.  Thank you for the feedback.  This makes me feel better about doing the testing and additional work to migrate to Distributed switching and Nexus 1000v.

[slabaky]

To add some additional comments, with Nexus 1000V Network admin can define network policies on the virtual supervisor module such as QoS, VLAN, ACL… these policies appear on the VM vcenter server as a port-group that can be applied to individual VMs. with Nexus 1000V you are able to independently configure the VM network via ESX or NX-OS. When polices are attached to a Virtual machine they follow the VM throughout the lifecycle, when VMotion happens, VM can continue traffic monitoring such as netflow and ERSPAN because the Nexus 1000V preserves the entire VM state.

Additionally the N1K supports a variety of virtualisation of layer 4 to 7 such as Virtual gateway
for example traffic destined to a VM that requires firewall services, instead of installing firewall services on each physical server, Cisco Vpath automatically redirects the traffic to another Virtual server that has the firewall VM and provides a better deployment model, additionally the N1K supports decision cashing on the VEM, in this case the VSM cashes the firewall decision in a flow table, like this future packets no need to be sent to the external server, next traffic will take place in the VEM. Vpath can be applied to different services not only VSG

[syedal3]

Hi,

Following link will be helpful

http://www.cisco.com/c/en/us/products/collateral/switches/nexus-1000v-switch-vmware-vsphere/data_sheet_c78-492971.html

Regards,

Syed