With all the turmoil around VMware, should you consider moving to an alternative platform? With all the changes happening in VMware, both at the company and licensing changes many are looking towards changing to another virtualization layer. Why?
- 1: They have gotten a price increase on their license, for many of the smaller customers they will most likely get a price increase in their license. This is because of the product bundles that are changing (VMware by Broadcom Dramatically Simplifies Offer Lineup and Licensing Model – VMware News and Stories) which makes sense for the larger enterprises that are using more of the stack, but for the smaller customers that are only using vSphere well the price jump can be quite significant.
While the price increase is different from customer to customer, you can look at this Reddit thread and see for yourself how different companies are impacted (2024 Prices! Did they increase for you? : vmware (reddit.com) and this is just one example.
- 2: The EUC part of VMware/Broadcom is now a separate company and has been acquired by a private equity fund named KKR, meaning that while it means that they will be able to decide their own strategy and roadmap moving forward, it also means that it doesn’t make sense for them to continue investing so heavily on integrating with vSphere. (KKR’s $3.8B Rescue Of Broadcom’s VMware EUC Business: 5 Things To Know (crn.com))
- 3: Many key people leaving the company. While this is not going to impact how companies run VMware today, it will impact development and strategy in key areas for Broadcom/VMware moving forward.
Should you move to another virtualization stack? This is something that I have discussed with a lot of customers lately. Some of them have already made up their minds in terms of which direction they want to go, and some are asking for guidance on what are important things to consider if they decide to move to another virtualization platform for their workloads. Let us look at some important considerations.
1: Knowledge and expertise – Moving from one virtualization layer to another, while technically can be quite easy it also means that you need to build up new knowledge on another set of technology. This is not something that is done overnight and needs to be planned properly. Secondly depending on which technology you choose, the pool of people that know that specific technology might also be a lot fewer then the ones that know VMware.
2: Integrations or dependencies with the virtualization layer – While not something that should impact the decision of moving or staying on a virtualization platform it is something that needs to be considered. You might not think about it, but several technologies integrate directly with the different vSphere APIs
- Backup products (such as Veeam)
- VDI services such as Horizon and Citrix
- Storage services (such as CSI drivers for Kubernetes)
- Networking products which are running as virtual appliances (F5, NetScaler, Cisco, Palo Alto)
- Security products such as Trend have integrations into network APIs.
- Monitoring tools. While some can standardize on VMware monitoring tools you need to consider what kind of alternative can be done here.
You might also have products or features that you are using on top such as Kubernetes platform (for instance Tanzu from VMware) which depending on which version does not directly support other hypervisors, this also requires you to change Kubernetes platform as well if you switch to another virtualization platform. So that means that while changing to another virtualization layer, might also require you do change those 3.party tools if they do not support the new platform.
3: Then you also need to consider the technology stack, for instance, if you are using vSphere with VSAN (Software-defined storage) NSX (Software-defined networking). Moving to another virtualization layer means that these features will not be available with using 3 party alternatives. For instance might be that you are using VMware vDS with Multicast, which is not support on all the alternatives.
Some vendors provide alternative technologies such as Nutanix and Microsoft. However, the SDN capabilities from Microsoft are not as feature rich and easily manageable compared to VMware NSX.
You can see here some examples on how to configure Load Balancing using the Azure Stack SDN stack for instance Manage Software Load Balancer for SDN – Azure Stack HCI | Microsoft Learn
if you are just using VMware with vSphere to have a place to run your virtual machines, well then this point is not that relevant.
4: However another aspect is the hardware that you have, if you plan to move to another virtualization layer you also need to consider what your hardware supports. Most of modern servers support most virtualization platforms, but you also need to consider network drivers, RAID controllers, underlying storage, and other hardware components that you have like GPU cards.
For instance, if you have a virtualization stack that uses GPUs for VDI environments for vGPU, well then you need to have a similar platform that provides the same capabilities. Microsoft for instance supports a feature called GPU-P Partition and share GPUs with Azure Stack HCI virtual machines – Azure Stack HCI | Microsoft Learn Nutanix also supports vGPU (NVIDIA vGPU on Nutanix) same with Citrix XenServer.
You might also have other requirements, that the platform should be air-gapped meaning that you cannot use public cloud IaaS services as an alternative. You might also need specific high-availability scenarios such as what VMware has had with fault tolerance which is not directly available on most of the alternatives.
5: Support for workloads. In many cases you might be running virtual workloads such as Oracle RAC on Linux, MSSQL Clusters, Exchange and if you want to run this virtually it also need to be supported by the vendors.
5: The final part is of course the cost related to the new platform, while VMware can be expensive on its own, moving to another platform is not just about the migrating the virtual machines, but there are multiple things we need to consider in terms of cost. 1: The product with licenses (and additional hardware if needed) 2: Ecosystem tools (might be that we need a new backup solution or replace other components) 3: Support 4: Training of the staff to manage the new platform and also 5: The migration project to move resources from the existing virtualization platform to the new one.
So, what options do we have? Well it is a pretty long list of alternatives that provide a virtualization platform but not all of them provide all the addon capabilities like NSX and VSAN, but ill get back to those parts in a moment.
From Microsoft, we have Microsoft Hyper-V which can run on any Windows Server (that is certified for Windows Server) and can be used with a regular SAN or with S2D (Storage Spaces Direct). Or you have Azure Stack HCI which uses the same capabilities as Hyper-V with S2D but also uses cloud-based management. Azure Stack HCI while running Hyper-V underneath also has some additional features that are not available in the standalone Windows server w/Hyper-V such as Stretched Cluster Compare Azure Stack HCI to Windows Server – Azure Stack HCI | Microsoft Learn. Microsoft recently released a new version (which you can read more about the new features here –> What’s new in Azure Stack HCI, version 23H2 release – Azure Stack HCI | Microsoft Learn)
Then we have Nutanix, which has been one of the few real competitors to VMware SDDC. Back when Gartner still made their magic quadrant for HCI it was always VMware and Nutanix in the leaders quadrant.
After the latest release in 2021, Gartner stopped making the HCI quadrant and started making the Hybrid Platforms quadrant instead. Which of course chages the layout a bit.
Nutanix also has some SDN capabilities with their Flow product. Also has decent support from different backup vendors, but it is not straight forward to get Nutanix running on existing hardware. Also if you need to have it conneting to lets say a SAN you cant connect it directly to their hypervisor. While they offer storage nodes you have to have it within their fabric. Nutanix has their own hypervisor called AHV (Which is built in top of KVM) but managed from their management product.
Red Hat Virtualization could of course be an option, but Red Hat decided to discontinue it in 2022 (which will be EoS later this year) On August 31, 2022, Red Hat Virtualization enters the Maintenance Support Phase – Red Hat Customer Portal
You could use KVM, which is open-source but it has limited integrations and support from backup vendors and features compared. Lastly you do not have any support options available, however it can be bought from partners depending on where you are located. For instance AHV and AWS have been using KVM for a while, so the product is pretty solid.
What about Citrix XenServer? XenServer has decent support for hardware, support different GPU configurations. Unfortunately not many of the large backup vendors support XenServer which means that you would need to use tools like Xen-Orchestra to do backup of the VMs (it also provides a better management plane on top) XenServer does not provide a software-defined storage service like VSAN and also none SDN capabilities. The issue is also having tools that can do backup of virtual machines on XenServer, there are a few tools that do it, but in most cases you would need to change to another vendor. So the ecosystem is a lot smaller compared to the others.
Then you have Proxmox which is built on top of KVM, also open-source tooling. Proxmox also has its own backup tool and can also be bought with support from them directly. Also Proxmox does not natively have capabilities like software-defined storage, so in case you would want that, you would need to set up a seperate service for that such as Ceph.
The final option is product called Kubevirt, which is essentially managing VMs from Kubernetes. In Kubevirt, virtual machines (VMs) are deployed and managed within Kubernetes, but they are not directly encapsulated within pods. Instead, Kubevirt leverages the concept of Custom Resources and Controllers to provide VM orchestration capabilities.
We have also Harvester which is a product from the Rancher team which is aimed at being an open-source HCI solution, which combines Kubernetes + Longhorn (Software-defined storage) and Kubevirt, as seen in the screenshot below.
Now the issue here is the ecosystem, while small there are a few vendors that support backup of VMs running on Kubevirt, such as Portworx. Kubevirt also supports features like PCI-E hardware passtrough and vGPU, meaning it allows us to map a hardware device such as a GPU directly into a virtual machine.
These are some of the virtualization platforms that are available in the market, if you plan to move away from VMware as your virtualization layer. Ufortunately there are no right answers to which direction you should take. It is dependent on your requirements, direction, strategy, knowledge and expertise but also cost. For some the right answer will still be to use VMware, since there is no tech stack that can match what their current technology stack. For some the right answer might be public cloud or one of these alternatives.