In the previous chapters, we not only introduce the isolation detection when the isolation response is triggered, but also introduce the effect of false positives. When the quarantine response is set to power off or shutdown, the host becomes isolated and HA restarts the virtual machine on the host. However, this also means that there is no proper redundancy. Isolated responses may not be necessary to trigger, and downtime should be prevented.
To improve the flexibility of the network, VMware designed the concept of port groups for Vmkernel and virtual machines at the virtualization layer, which is particularly important for managing networks when discussing ha.
Reference
A port group is a single logical network card consisting of several physical network adapters that can be used for network fault tolerance and load balancing.
With this mechanism, it can increase the redundancy of the management network, reduce the chance of an isolated event, and, of course, provide redundancy for other port groups, but this is not the subject of this chapter. Another option is to configure an additional management network to enable the management network on another Vmkernel port, with the fact that if multiple vmkernel are on the same subnet, HA will use all ports to manage traffic, even if only one is specified for administration!
Although there are many configurations and supports, we recommend a simple but highly resilient configuration. In our example will include vmotion, when the integration of Vmotion and management network into a virtual switch is a very common configuration, but also the best practice in the industry.
Demand
2 Physical network adapters
VLAN trunking
Suggestions
2 Physical Switches
If you can, turn on link state tracking to ensure a report is issued when a link fails
The virtual switch is configured as follows
VSwitch 0:2 Physical network adapters (VMNIC0 and Vmnic 1)
2 port groups (Management network and Vmotion)
Management Network Vmnic 0 active/vmnic 1 Standy
VMotion vmnic 1 Active/vmnic 0 Standy
Failure recovery setting is no
Each port group is labeled with a VLAN ID and runs on its own proprietary network card. Only in the event of a failure to switch to the standby network card, we strongly recommend that failback set to No, so as to avoid unnecessary orphaned events, when the physical switch is started when the path does not have traffic, but the port group is reported as "up", This can cause orphaned events.
Advantages:
The management network and vmotion require only 2 NICs, which are particularly important in the blade environment and are easy to configure
Disadvantages
Only a single active path of the heartbeat
Next, describe the Active/standy scenario.
Figure 24: Active-standy Design for the management network
In order to increase flexibility, we also recommend the implementation of the advanced option to set the network card port using a different PCI bus-preferably different brands of different models of the network card, when using a different network card, can even reduce the driver failure of the NIC.