On September 15th, VMware has announced new versions of vSphere, vSAN and VMware Cloud Foundation (VCF) and I must say that I am pretty impressed with the functionality changes.

In a quick overview, the main changes are:

• vSphere 7.0U1
  • vSphere with integrated Kubernetes
  • Melvin has grown
  • Cluster size increase
  • EVC for Graphics
  • LifeCycle Manager enhancements
  • vCenter Connect
  • vSphere Ideas
• vSAN 7.0U1
  • HCI Mesh
  • File Servics: SMB support!
  • Shared Witness
  • vLCM Enhancements
• VCF 4.1
  • NSX-T 3.0.1
  • VCF Remote Clusters
  • Parellel NSX-T Upgrades

First of all vSphere. In this new version, we can leverage the Kubernetes integration that was previously only available in combination with VCF. But from now on, it can also be added to “normal” vSphere implementations, without the need to use NSX as a network virtualization product. You will need distributed switches and there are some requirements there, but definitely an easier implementation than a full-blown NSX-T deployment:

Although I am very fond of NSX, this gives a lot of customers the chance to use vSphere with integrated Kubernetes and that is definitely a good thing.

Second change, for those who recognize the joke:

(source: https://twitter.com/melvinvmmonster)

Melvin was the name given to the Monster VM, and from the start of virtualization has always been a cult-figure. In the last versions of vSphere there have been enhancements to the largest possible VM configuration, but there wasn’t a lot of noise about it. This release Melvin has definitely had a growth spurt. He has grown to the amazing size of:

• 24 TB of memory (from 6 TB)

Okay, that is not something I expect to come across “in the wild”, but there must be some use cases that want this :).

The largest cluster size has also grown, to 96 nodes, but that is without vSAN. The vSAN maximum cluster size is still the same (64). The maximum number of VM’s in a cluster has grown to 10K.

A very interesting new feature is EVC for Graphics. Where EVC has greatly enhanced the flexibility around mixed CPU clusters, this is a way to grow clusters with different hardware (getting ready for the future), but without loosing flexibility. So you can buy new hardware, let it run on the same EVC level as the older hardware, while phasing out the older hardware:

New functionality in LifeCycle Manager is (among others) the integration with NSX-T (starting with the upcoming NSX-T release):

vSAN 7.0U1 also comes with a lot of new features. As was said during the BEAP (Blogger Early Access Program) session is that a lot of features might not be very flashy, but they are extremely useful when it comes to the enhancement of the product. Lot of improvements on manageability and performance.

Apart from those improvements, a couple of new features that are very visible:

HCI Mesh gives the ability to connect vSAN Clusters to other vSAN clusters and use each others datastore. And all this done through vSAN’s native protocol, which optimizes efficiency and simplicity. This gives (among others) the ability to do a vMotion, without moving the storage, when moving a VM from one cluster to another. A very interesting feature, in my opinion:

Another big announcement is the addition of SMB to the list of File Services:

When designing vSAN environments I often have a debate around using deduplication and compression because it is a combined feature. One of the consequences of using this feature is that deduplication means that a whole disk group is seen as a single entity. When one device (cache or capacity) within that entity fails, the whole disk group fails.

With the new version, it is possible to use compression only, without also using deduplication. That reduces the risk, while still getting optimizations on physical device usage.

When designing clusters, I always get into discussions about slack space. 25-30% is the desired value, in order to move data around or respond to failures of disk groups or hosts. With larger clusters, 25-30% is a very big chunk of usable data. So in 7.0U1, VMware has changed this and are now enhancing usable capacity, by using two kinds of “reservations”. One reservation is called “Host rebuild reserve“, which means the amount of space necessary to rebuild a complete host, when failure happens, the other reservations is called “Operations reserve” which is used to perform operational tasks.

The total amount of required space for these actions is based on the size of the cluster. Larger cluster, means a lower percentage:

and this is also shown in the overview of the used capacity within a cluster:

For some environments, with a lot of small clusters, across multiple sites, maintaining Witness hosts for all those clusters, could become a hassle. With the new version comes the ability to combine those roles into one witness appliance:

From a security perspective, enhancements have been made around data in transit. It is now possible to encrypt this traffic. And to utilize this, there is no need for a Key Management Server (KMS):

I’m not going to name all the enhancements “under the hood”, but one that I am very excited about is the change in the boot-sequence of a vSAN host. I have spent days (in total ;)) waiting for a cache rebuild when a vSAN host rebooted, but from now on, that is no longer the case. The process of rebooting a vSAN host has been improved, to be smart and optimize this process. Very happy about that :).

VMware Cloud Foundation 4.1 comes with vSphere 7.0U1 and vSAN 7.0U1, but there is more.

It also comes with NSX-T 3.0.2 (and a new blog on this might be following soon ;)).

Another addition to the functionality of VCF is the ability to do VCF Remote Cluster Deployment. Targeted to small environments (minimum 3, maximum 4 nodes per cluster), to be managed from a central location:

Apart from this, some other interesting new features, like the addition of vVOLs as primary storage and NSX-T parallel upgrades.

All in all a lot to look forward to! When? Well, let’s say I hope very soon :).