cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
4064
Views
0
Helpful
1
Replies

UCS vNIC Connection Policies

Fred Weers
Level 1
Level 1

Hi all,

Searching the web does not give me the information I seek so I come to you.

We have a new UCS cluster, and now I noticed the selection between Connection Policies when creating a vNIC Template.

I'm trying to understand the real difference between and can I get any benefit when using the one or the other.

Dynamic vNIC
usNIC
VMQ

We run primarily vmware and oracle on the ucs blades.
Any insight regarding performance support and redundancy on these platforms are more than welcome.

Thanks in advanced

Fred

1 Accepted Solution

Accepted Solutions

Walter Dey
VIP Alumni
VIP Alumni

Hi Fred

Regarding dynamic vnic's

The Virtual Machines using the VMDirectPath Dynamic vNICs should have memory reserved

VM-FEX VMDirectPath is a feature where the network IO bypasses the Hypervisor’s network kernel and communicates directly with the Cisco VIC adapter.
This helps in offloading the host CPU and Memory cycles to handle VM’s networks.

In the VM-FEX Port Profile, you need to configure the Dynamic Ethernet interface’s actual VLAN, QOS, pinning etc.
Unlike static vNICs, Dynamic vNICs are not configured upfront.

http://www.cisco.com/c/en/us/solutions/collateral/data-center-virtualization/unified-computing/vm_fex_best_practices_deployment_guide.html

In summary: VM-FEX applies essentially to VM's bypassing the hypervisor. VM-fex supports the following hypervisors: VMware, KVM, Hyper-V.

Regarding usNIC

http://www.cisco.com/c/en/us/td/docs/unified_computing/ucs/usnic/b/deployment/3-1/b_Cisco_usNIC_Deployment_Guide_For_B-Series_Blade_Servers.html

The Cisco user-space NIC (Cisco usNIC) feature improves the performance of software applications that run on the Cisco UCS servers in your data center by bypassing the kernel when sending and receiving networking packets. The applications interact directly with a Cisco UCS VIC second generation or later adapter, which improves the networking performance of your high-performance computing cluster. To benefit from Cisco usNIC, your applications must use the Message Passing Interface (MPI) or Libfabric interface instead of sockets or other communication APIs.

To benefit from Cisco usNIC, your configuration has the following prerequisites:

  • A supported Linux operating system distribution release. For more information on supported Linux operating system releases, please refer to the UCS Hardware and Software Compatibility Tool.

  • The UCS Driver ISO corresponding to the UCS server model, selected Linux operating system, and version of UCS firmware installed on the server as identified by the UCS Hardware and Software Compatibility Tool. For more information, see Downloading Cisco UCS VIC drivers.

  • A supported MPI implementation, such as IBM Sprectrum MPI, the open source Community Open MPI package, or version 4 or 5 of the Intel ® MPI Library. If the Intel ® MPI Library is used, the network must be configured with flow control enabled.

View solution in original post

1 Reply 1

Walter Dey
VIP Alumni
VIP Alumni

Hi Fred

Regarding dynamic vnic's

The Virtual Machines using the VMDirectPath Dynamic vNICs should have memory reserved

VM-FEX VMDirectPath is a feature where the network IO bypasses the Hypervisor’s network kernel and communicates directly with the Cisco VIC adapter.
This helps in offloading the host CPU and Memory cycles to handle VM’s networks.

In the VM-FEX Port Profile, you need to configure the Dynamic Ethernet interface’s actual VLAN, QOS, pinning etc.
Unlike static vNICs, Dynamic vNICs are not configured upfront.

http://www.cisco.com/c/en/us/solutions/collateral/data-center-virtualization/unified-computing/vm_fex_best_practices_deployment_guide.html

In summary: VM-FEX applies essentially to VM's bypassing the hypervisor. VM-fex supports the following hypervisors: VMware, KVM, Hyper-V.

Regarding usNIC

http://www.cisco.com/c/en/us/td/docs/unified_computing/ucs/usnic/b/deployment/3-1/b_Cisco_usNIC_Deployment_Guide_For_B-Series_Blade_Servers.html

The Cisco user-space NIC (Cisco usNIC) feature improves the performance of software applications that run on the Cisco UCS servers in your data center by bypassing the kernel when sending and receiving networking packets. The applications interact directly with a Cisco UCS VIC second generation or later adapter, which improves the networking performance of your high-performance computing cluster. To benefit from Cisco usNIC, your applications must use the Message Passing Interface (MPI) or Libfabric interface instead of sockets or other communication APIs.

To benefit from Cisco usNIC, your configuration has the following prerequisites:

  • A supported Linux operating system distribution release. For more information on supported Linux operating system releases, please refer to the UCS Hardware and Software Compatibility Tool.

  • The UCS Driver ISO corresponding to the UCS server model, selected Linux operating system, and version of UCS firmware installed on the server as identified by the UCS Hardware and Software Compatibility Tool. For more information, see Downloading Cisco UCS VIC drivers.

  • A supported MPI implementation, such as IBM Sprectrum MPI, the open source Community Open MPI package, or version 4 or 5 of the Intel ® MPI Library. If the Intel ® MPI Library is used, the network must be configured with flow control enabled.

Review Cisco Networking for a $25 gift card

Review Cisco Networking for a $25 gift card