Looking to transform your data center into more modern, easy to manage and cloud-driven infrastructure? Well, first of all – congratulations! You’re on the right path 😀 – Azure Stack HCI is probably the hottest cake in the hybrid cloud market right now. If you’re planning to move onto Azure Stack HCI but not sure what it would take or what you’d need? No fear – Microsoft has a handy sizing tool that will help you estimate what kind of infrastructure you’d need and what things to consider to make this shift. You can find it here:
Azure Stack HCI Solutions | Microsoft (azurestackhci-webapplication.azurewebsites.net) – you will need a Microsoft account to access this.
Create a solution.
Choose the system type. Integrated systems like Dell Integrated Solutions for Azure Stack HCI are the end-to-end solutions extensively tested and maintained by the OEM partner. It includes benefits like full deployment and maintenance support, integrated support between Dell and Microsoft, continuous updates, and well tested, purpose-built hardware to name a few. My colleague Jayanth has laid out the differences in great detail on his YouTube channel – Jayanth On Cloud – here.
Choose the CPU manufacturer – AMD or Intel.
Choose your preferred solution partner to build your integrated HCI solution. You will notice there are some options greyed out, which means they are simply validated nodes and not the integrated ones.
Select the redundancy levels you want in your cluster in terms of number of nodes and storage drives. This ties into S2D that AzS HCI uses at its SDS level.
Obvious thought would be to go for as much resiliency as you can, but keep in mind this will result in increased cost (though it may be well worth it in a long run).
Next, choose the storage type amongst All NVMe (All NVMe for both capacity and cache), All flash (all SSD for both capacity and cache or SSD for capacity + NVMe for cache), and Hybrid (HDD for capacity + SSD for cache). I will choose hybrid in this case.
Choose your preference for whether you want better performance or more capacity. This will determine what kind of mirroring will be used (2-way with 50% capacity or 3-way with 33.3% capacity etc.) This is of course a relative question but in vast cases it is safer to assume that if you’re going for hybrid storage you want to utilize HDD for capacity. If the applications you’re planning to use are performance sensitive, my general advice would be to go all flash (SSD/NVMe).
Next, if you have plans for future growth, it is crucial that you have the capacity available. To plug in more RAM/storage, you’d need to have that reserve to them on later. Also, if you have plans for growth, start from a minimum of 3 nodes, so you can easily accommodate more nodes without any drastic measures.
You can choose to use SDN too, but I’ll leave that unchecked for now.
Now we can move on to what kind of workloads we’re planning to run on our AzS HCI cluster.
Go ahead and Add workload.
You will be presented with this wizard. Workload type can be anything specific like SQL and VDI, or you can also specify something more generic as General. Note that what you choose here does not have any actual impact on what you can run on your cluster, this is only an estimate for capacity/compute you might need to run those workloads.
I’ll choose some random inputs here. I’ll go and repeat this a couple more times so we can see what it looks like.
I went ahead and added a SQL and a VDI workloads and at the end this is what I see:
So this turns out to be our final-ish requirement. Now we can go ahead and see the results!
And this is what it’s suggesting us:
A Dell AX740xd 2-node cluster should do our job. Note that the values required and suggested are not exact matches, because you have to factor in resiliency/future growth and availability.
Glancing through the solution, this actually looks pretty good. However, keep in mind that this does not include any networking devices such as racks/switches/cables etc. So those will be added on later.
Also, treat this as reference only. There is also a Dell version of HCI sizer which factors in for a lot more than this. Dell integrated system for AzS HCI being an integrated system also adds on other services such as end-to-end deployment/unified support and day N troubleshooting assistance. Also Dell will have it’s own recommendations and best practices based on their field experiences and rigorous validation they do. For example, like I said previously – although the Microsoft sizer says 2 nodes is enough, a Dell engineer will recommend starting with a 3-node cluster instead so we can scale with ease beyond that.
This sizer is also only for AzS HCI, not Windows Server HCI. So if you want to know what EXACTLY your solution should look like, get in touch with Dell!