vSphere 6.0 Announced

vSphere 6.0 is now available and there are some great new enhancements with the new version.  Here are some of the many highlights from today’s announcement.

 Speeds and Feeds

As with the new version of anything things are bigger and faster.  vSphere 6.0 is no exception.

• 64 hosts per cluster, up from 32
• 8000 Virtual Machines per Cluster, up from 4000
• 480 CPUs, up from 320 CPUs
• 12 TB RAM, up from 4 TB (if someone has 12 TB of RAM in a box, please let me know how long it takes to do a memory check.  vSphere 7 might be out by then.
• 1000 Virtual Machines per host, up from 512 Virtual Machines

Virtual Machine Hardware version 11 allows for:

• 128 vCPUs
• Hot-add RAM now is vNUMA aware
• Serial Ports now have a maximum of 32 ( I KNOW CAN YOU BELIEVE HOW AWESOME THIS IS!!!!!)
• vMotion is supported for Cluster Across Boxes with physical compatibility mode RDMs

vCenter 6 now has parity between the Windows version and the appliance version.  If you use the appliance though, your database options are still limited to vPostgres or Oracle.

• 1000 VMs per vCenter
• 10,000 Powered-On VMs per vCenter
• 64 Hosts per Cluster
• 8,000 VMs per Cluster

Web Client is drastically improved.  There have been plenty of grumblings about the vSphere web client in the past, but it looks like we should give the new client a go.  Mainly because it will be a necessary piece of the infrastructure for some time to come.  In any case, try using it with an open mind.  Performance has been dramatically improved,  and the layout has been adjusted to look more like the C# client.  I think we’ll like it.

Fault Tolerance

Fault Tolerance is nothing new, but there has been a big wait for vSMP for FT.  I mean if you’ve got a VM that’s so important that you need two of them in lockstep, what are the odds that this VM only have one processor?  Usually, really important VMs also have high performance characteristics.  Well, relax, you can now protect up to 4 vCPUs in a single VM, but pay close attention to the caveats.

• Fault Tolerance is now able to be hot configured
• Protect 4 vCPUs
• Allows snapshots now
• allows paravirtual devices
• Works with SRM, and VDS

Caveats

• Requires an additional vmdk ( This makes storage fault tolerant as well which requires twice as much storage, but also adds to availability)
• Requires 10 GbE
• Max of 4 protected virtual machines

Introducing the Platform Service Controller

There is a new construct in town and it’s the Platform Services Controller (PSC).  This controller manages Single Sign-On (SSO), Licensing and Certificate Authority.  The PSC will link all of your vCenters together automatically if they are in the same SSO domain.  You can imagine this controller being the new way to manage highly available vCenters in the future.

Also, The PSC is a certificate authority.  All of those pesky warnings about untrusted certificates are soon a thing of the past.  And no more grueling through updating certificates on your hosts and services.   Now certificates will be as easy as right clicking and asking for a new one.   You can make the PSC an intermediate CA if you’ve got an existing PKI.

Big Changes for vMotion

There are quite a few changes to vMotion now and I think we’ll be seeing a lot of new uses for the technology.

• vMotion between switches.  (This requires Layer 2 connectivity.  It will not re-IP the machine, but you can now migrate between switches if you need)
• Migrating between vDS switches keeps the metadata for historical purposes.
• You can now vMotion between vCenters.  This may be great for migrations in the future.
• Long Distance vMotion as long as RTT is less than 100ms
• vMotion over Layer 3.  Now we don’t even need to have layer 2 adjacent hosts to vMotion.  We can now route to do a vMotion.