So the worst kept secret in Virtualisation has now been made official, vSphere 6 has been announced with Pat Gelsinger keen to emphasise that this is the biggest release yet with 650 new features. Sounds good, but just what is being added and what does it mean for your Virtual Estate? Here’s a summary of the key enhancements and why you might just get a little excited….
So first up those much-loved configuration maximums are changing up yet again and there are no mean portions here with double the number of hosts per cluster and nearly double the amount of VMs per host. Not bad. More explicitly, each vSphere 6 host will support:
- 480 Physical CPUs per Host (up from 320 in 5.5)
- 12TB Physical RAM (tripled from a miserly 4TB in 5.5)
- 1000 VMs per host (nearly doubled from 512 in 5.5)
- 64 Hosts per Cluster (doubled from 32 in 5.5)
Obviously this means that for those of us lucky enough to have the need your clusters can be so much larger and your hosts loaded up with borderline ridiculous amounts of compute resource. Why would you need so much compute off one host you may ask? Well with vSphere 6 comes Virtual hardware v11 for your Virtual Machines, which gives you this in a VM:
- 128 vCPUs (double the 64 vCPUs allowed in 5.5)
- 4TB RAM (up from just 1TB in 5.5)
- Hot-add RAM is now vNUMA aware
- xHCI 1.0 controller compatible with OS X 10.8+
Okay so you can have some ENORMOUS VMs, this means that applications like SAP HANA are no longer just a dream for Virtualisation enthusiasts as the new supported maximums will allow the full virtualisation of your SAP HANA suite. Sweet.
It doesn’t stop there either. It wasn’t too long ago that the vCenter Server Appliance was the domain of small-scale dev environments and labs, then 5.5 came along and ramped up the supported hosts to something more usable. vSphere 6 then, will give you up to 1000 hosts and 10000 VMs from your vCSA, Windows-aphobes rejoice.
So that’s what’s getting all the press, some other nice little features are also being slipped in to vSphere 6, like the ability to run ESXCLI commands against vCenter and not just a host directly, and also better local host account and password management.
Fault Tolerance – SMP FT is finally HERE!
Yes you’re not dreaming, at long last what VM admins have been crying out for since vTime began has arrived with vSphere 6- FT is finally a realistic option for zero-downtime availability for any Guest OS, as SMP FT allows you to configure a fault tolerant VM (allows you to HOT configure a fault tolerant VM no less) with up to 4 vCPUs! Not only that, SMP FT will work with any Virtual Disks AND supports snapshots.
Fault Tolerance is finally a viable solution for (according to VMware) 95% of Virtualised applications, including vCenter Server, albeit for small scale environments.
FT has been completely rewritten and I fully expect these new maximum supported values to go up and up with each new release, watch this space.
One of the highlights of the new feature list for me is the enhancements to vMotion. Prior to vSphere 6 your vCenter Server would serve as your boundary for any vMotion activity but no longer, vSphere 6 will allow you migrate live running VMs seamlessly from a host managed by one vCenter Server, to a host managed by a different vCenter Server- provided joint management and consistent L2 network connectivity between the two.
Couple this with the second major vMotion enhancement and the DR, DA and general ease of maintenance opportunities are staggering- vSphere 6 will also allow long distance vMotion. Provided latency remains below 100ms, you will be able to vMotion anywhere in the world (VMware are not defining a diastase maximum, it’s all down to the infrastructure), and L3 vMotion networking support will make this a lot easier to achieve.
This is great stuff, think about it: Is your DC threatened by fire, flood or server-eating dragon? vMotion it somewhere else. Done.
The feature-formerly-known-as Project Fargo, VMFork or a myriad of other names (legend has it VMware named it something new for each tech partner working on the project so they could source any leaks that appeared. Sneaky buggers that they are). Instant clone is also something special, using Copy-On-Write you can now instant clone a VM in a matter of seconds- they will share disk and RAM space until the cloned VM modifies either at which point it’ll get it’s own delta, also meaning very small footprint. VMware reckon you can create 64 instant cloned VMs in under 6 seconds and time to provision a Virtual Desktop has been improved 13 fold, which is impressive by anyone’s standard and I for one can’t wait to see how this performs in the real world.
VMware Virtual Volumes
vSphere 6 also introduces VVOLS, essentially allowing non-VSAN adopters to get all the joy of Object-based storage from their traditional (and fully compliant, fully firmware’d up) SANs. What this means is that rather than the LUN being the unit of consumption and a hard-barrier to performance or resiliency requirements, VM storage is policy-based- multiple VMs with different mirror and striping requirements can now reside next to each other on the same LUN through VVOLS, so you don’t need to create a new LUN if you need a RAID6 for example, just configure it in the policy and apply. Done, and very handy in my opinion. Incidentally, vSAN itself isn’t excluded from awesome new features, look forward to an upcoming post for those!
So how’s that for starters? Excited about vSphere 6 yet?? As with all new VMware releases it’s a brave few who jump in immediately to upgrade their production environments and it will be interesting to see what (if any- best give VMware the benefit of the doubt here) issues come out of these new features, but there’s a lot here to get your teeth into and vSphere 6 could have a huge positive impact on your vKit.
For more details on these and any number of other features I didn’t have the time to document here, check out the official press release here.
Until next time, thanks for reading and let me know if you’ve had your hands on these new features yet and what you think.