Typical HV node will have 2 x 10 Core CPUs, 128 or 256GB of Ram, 5 x 960GB SSDs in raid 6 or 4 x 1.92TB drives in Raid 6. We use VMM for day to day management of the stand alone hyperV nodes. And this is only for 2 product categories, shared hosting and enterprise cloud (dirty word I know!) Just to say that we have 100s of HyperV nodes and approx 1000 VMs under management. We also host around 100k MySQL dbs on around 150VMs across a number of nodes.
We have a bunch of S2D clusters with 100+ VMs in each one and then we have around 100 HyperV nodes that we use for Windows and Linux shared hosting. In my "place of work" we have MANY many HyperV nodes. I realize this is a lot, but interested to hear from others, or even be pointed to supplemental resources such as customer transition case study videos, etc.
How are you handling backups? I have used Veeam and Cohesity for larger scale deployments. If you do not use SCOM in your environment can you recommend other monitoring platforms worth consideration that contain a solid Hyper-V focus? I have used Solarwinds Virtualization Manager, Foglight for Virtualization, Veeam Monitor, SCOM with Veeam Management Pack, and LogicMonitor. What is your stance on forgoing SCVMM in favour of alternative tools such as 5nine's Cloud Manager? My biggest concern there is in the SDN support. Main challenges faced on a day-to-day basis and some of the approaches you have taken to mitigate those challenges by way of tools or processes. I am interested to hear your experiences on the overall stability of your HV environments of this size. My experience however is in larger distributed VMware environments, typically into the thousands of VMs. I am well versed in HV environments of up to 150 VMs utilizing SCVMM. regularly creating and decommissioning VMs, larger sized VMs, an increased requirement for high availability and service levels, increased management complexity, larger host pool to maintain and keep patched, etc.
Reason being, the additional complexity that comes with day-to-day management, the increased likelihood of the environment being more fluid and dynamic - IE. I am looking to hear from customers that are running in the 200+ VM range. My experience is the average HV environment is 2-4 hosts and 30-50 VMs. I am interested in hearing from other environments that have a larger than average Hyper-V environment. I do not necessarily agree with the direction but that is besides the point. Self-service for developers, ASR and the opportunity to forgo future licensing and support costs are some of the drivers.
There is a large driver towards Azure, with a requirement to maintain an on-premises presence. There are multiple factors in the transition, some technical and some political.
The environment will continue to utilize the existing shared storage but will phase it out over time in favour of hyper-converged. The end-state will be Hyper-V 2019 on a 6 host cluster / site with a combination of shared storage (3PAR and Nimble) and S2D being the future. Current estate is 16 ESXi v6 hosts, 490 VMs, 2 geographically separated data centres. I am currently engaged with a company that is in the process of migrating from VMware to Hyper-V.