Uncovering VMware vSphere 5.1′s “Best Kept Secrets” Part 2


by Bill Oyler | 06.12.2013
Categories: Blog

So in my first article I revealed the first 3 of the 6 best secrets I have uncovered while using VMware’s vSphere 5.1. Today I will conclude the article with the last three.

#4:  Multiple NIC vMotion

Speaking of vMotion being around for a decade, have you noticed that VMware admins tend to complain about how “slow” vMotion performs when they put a host into Maintenance Mode?  The truth is that vMotion is remarkably fast and can efficiently saturate an Ethernet link (even approaching 10 GigE throughput).  But moving 256 GB worth of RAM from one host to another certainly can take a long time, especially at 1 GigE speeds.  One easy way to boost vMotion performance is by leveraging the new “Multiple NIC vMotion” feature introduced with vSphere 5.0.  You need to perform specific configuration tasks to enable this feature (you can’t just add multiple vmnics to a vSwitch).  Follow the below KB article and watch Duncan’s video to see how to set it up.  Once you do, you will wonder why you didn’t set this up sooner! 

Multiple-NIC vMotion in vSphere 5 (2007467) 
  • TIP: You can use up to (16) 1 GigE NICs or up to (4) 10 GigE NICs for Multiple NIC vMotion. 
  • TIP: Multiple NIC vMotion is included with all vSphere licenses that support vMotion.  

#5:  vSphere Fault Tolerance (VFT)

vSphere HA is a fantastic high availability solution, but virtual machines will still crash for 60+ seconds in the event of a physical server failure.  Typically it will take several minutes for a virtual machine to be restarted on another vSphere host using VMware HA.  While this is a suitable level of HA for 99+% of my customers’ applications, there still may be one or two “key” applications that cannot tolerate even 10 seconds of downtime.  If you have such an application, and it meets the requirements for vSphere Fault Tolerance (such as a single virtual CPU), Fault Tolerance is the perfect solution for you.  Fault Tolerance will essentially “mirror” the virtual machine on two vSphere hosts at the same time, so they are running in lockstep.  Every single keystroke, mouse movement, and network packet is mirrored to the other virtual machine running on the other physical server.  When a physical server crashes, the mirrored virtual machine takes over immediately, with “near zero” downtime. 

  • TIP: vSphere Fault Tolerance is included with the following vSphere licenses:
    • Standard
    • Enterprise
    • Enterprise Plus

#6:  Virtual Distributed Switch (VDS)

Are you tired of manually configuring vSwitches on each of your vSphere hosts?  Have you ever made a mistake configuring a vSwitch, only to discover the mistake at the worst possible time (such as after a vMotion)?  If so, you may want to consider using a Virtual Distributed Switch that spans your entire vSphere cluster.  The Virtual Distributed Switch (VDS) has been around since vSphere 4, but I find that it is often ignored, even in environments that are licensed for the VDS.

Enhancements to the VDS in vSphere 5.1 make it even more resilient and useful than it was in the past: 

  • Management Network Rollback and Recovery to correct mistakes that are made to the vSwitch.
  • Network Health Check to detect common network configuration issues, such as VLAN tagging mistakes, MTU mismatches, and NIC teaming/load balancing mismatches.
  • Support for RSPAN, ERSPAN, and Netflow.
  • Ability to backup and restore vSwitch configurations. 
  • TIP: Virtual Distributed Switch is included with the following vSphere license:
    • Enterprise Plus