In the Virtual Infrastructure 3 Online Library it is stated that "
To configure a NIC team for the service console, configure the vNICs in vSwitch configuration for Active/standby configuration
"
Can anybody explain to me why that is? it would seem more logical to put both nics on active/active to make use of the combined throughput.
thanks
Henkie
Active / passive is for pure redundancy. If you have your connection going in an active/active mode out of both interfaces, and your switch is not configured for that type of config, then there are scenarios in which the MAC address of the vm will show up on both ports on the switch at the same time. If you have default port security turned on, that's a no-no, and one, if not both ports will be shut off, causing connection failures on your interfaces. Also, if the teaming is sending out both interfaces, and the load balancing algorithms between the vSwitch and the pSwitch don't match, you'll get strange and/or intermittent connection faiures. Using active/passive avoids all these issues, and makes the troubleshooting much easier. The only time the 2nd interface will come up, is if the primary went down. You don't have to do any other config on the physical switch, and you don't have to figure out which NIC is failing in a team.
I agree though, that if you have the bandwidth, and are comfortable with the troubleshooting, then team your NICs. But, there is a level of understanding involved here as well, that everyone is not comfortable with, hence the active/passive would be easier to work with.
Also, remember, there is a common mis-conception of NIC teaming and link aggregation. If you have 2 NICs that are capable of 1 GB throughput, then you don't really get 2 GB. You double your total capable bandwidth, but no single packet will go at twice the speed, as you are still using 1 GB uplinks.
-KjB
Message was edited by: kjb007 : added bandwidth note
Don't know why it would say that...can you provide a link to the page where it is said and I'll see if I can get someone to take a look at it.
Ken Cline
Technical Director, Virtualization
VMware Communities User Moderator
http://pubs.vmware.com/vi35/wwhelp/wwhimpl/js/html/wwhelp.htm
Resource Management Guide
100. Setting Up neworking Redundancy
thnx
I think the goal of that recommendation is to help provide redundancy, and does not provide instructions on the added benefit of aggregation. It's showing you that you should have a redundant service console connection, whether it be a 2nd interface, or a 2nd teamed NIC.
I agree that it should also show an aggregate active/active config, but remember, that would need additional instruction on setting up port/ether channel on a physical switch, or instruction to use separate switches. Otherwise, you may be your connection bounce around. For instance, if you had a port group connected with 2 active NICs to a physical switch, then you may get intermittent connection failures, if the switch is not connected in a port channel. I just had this issue earlier this morning for a customer, as their vm could not ping its own gateway, and could not resolve the MAC address via arp of the router. A quick fix was to put the 2nd NIC in standby mode, and all was well with the world.
Hope that helps.
-KjB
You mean when you want inbound traffic to make use of the combined throughput, you need to do that, but we cannot configure the physical switches ourselves.
For outbound traffic it works fine
So i understand NIC teaming is possible with active/active, and when one nic fails your throughput is down to 1 GB ?
So what would be the use of active/passive assuming your switch accepts 2 GB? I would think you loose the standby nic! (until the other breaks down that is)
Henkie
Active / passive is for pure redundancy. If you have your connection going in an active/active mode out of both interfaces, and your switch is not configured for that type of config, then there are scenarios in which the MAC address of the vm will show up on both ports on the switch at the same time. If you have default port security turned on, that's a no-no, and one, if not both ports will be shut off, causing connection failures on your interfaces. Also, if the teaming is sending out both interfaces, and the load balancing algorithms between the vSwitch and the pSwitch don't match, you'll get strange and/or intermittent connection faiures. Using active/passive avoids all these issues, and makes the troubleshooting much easier. The only time the 2nd interface will come up, is if the primary went down. You don't have to do any other config on the physical switch, and you don't have to figure out which NIC is failing in a team.
I agree though, that if you have the bandwidth, and are comfortable with the troubleshooting, then team your NICs. But, there is a level of understanding involved here as well, that everyone is not comfortable with, hence the active/passive would be easier to work with.
Also, remember, there is a common mis-conception of NIC teaming and link aggregation. If you have 2 NICs that are capable of 1 GB throughput, then you don't really get 2 GB. You double your total capable bandwidth, but no single packet will go at twice the speed, as you are still using 1 GB uplinks.
-KjB
Message was edited by: kjb007 : added bandwidth note