A few weeks ago I was fortunate enough to attend VMware’s VMworld 2014 conference in San Francisco, CA. The show was simply amazing. I spent the show mostly meeting with the different technology vendors and other consulting groups, and feel that the relationships created at this event will last a lifetime.
With regards to database professionals worldwide, I feel that the most important announcement of this event for the upcoming release of vSphere 6.0 (no announced release date at this point) is the enhancement of multiple vCPU Fault Tolerance (SMP-FT). Previously, the fault tolerance feature within the vSphere suite was limited to one virtual CPU per VM. For SQL Server, this essentially prohibited the use of this feature. Now, vSphere 6.0 will have up to four vCPUs and 64GB of RAM available for use with VMs configured under VMware Fault Tolerance. I claim that this is one of the most significant developments of the year. Consider a SQL Server Availability Group configuration architected with each synchronous VM built with Fault Tolerance configured. Your most important business-critical database platforms, when constructed and managed properly, might actually achieve a 100% uptime for unplanned hardware failure outages.
Yes, the four vCPU and 64GB of RAM maximum is a limitation for DBAs, but I claim that a good number of business-critical SQL Server workloads could be made to fit within these constraints. Four is definitely better than one! It presents one more very viable option for building hardened and resilient platforms for your critical systems.
From a data professional perspective, the other announcements from VMworld are a bit less vital. For VMware administrators, the announcements are great and will save us all quite a lot of time for routine administration in the future.
- Cross-cluster and cross-vCenter vMotions are to be supported, which helps to improve a hardware upgrade by allowing you to move VMs between dissimilar clusters without downtime.
- A calculator called the vSphere Replication Calculator was announced that can help you determine the RPO, number of virtual machines, and the replication network bandwidth that is required to appropriately size a DR project.
- Performance of the vSphere Web Client is supposed to get better, part of which I have already addressed here for existing installations.
Otherwise, additional details on the upcoming release are pretty slim at this time, but I have high hopes for this next revision!
FYI – the vSphere 6.0 beta is open to everyone, but specific details of the implementation any other features not publicly announced are all under NDA, so keep this in mind if you explore this beta. I can’t wait to get this fired up in the lab so I can explore this upcoming release!