Back in May VMware announced its second generation of Data Center as a Service. Today I would like to take a look at what they announced and how does it fit with existing deployments.
If you haven’t heard about VMware Cloud on Dell EMC, this is a fully managed service running VMware SDDC on top of VxRail. I’ve always thought of this as the evolution of the vBlocks (VCE). However, in the old days, it was usually a partner or 3rd party company managing the environment, now it’s a mix of VMware and Dell support, being VMware the single point of contact.
So, what’s new?
Let’s start with the infrastructure. There is a new rack offering called R2, that provides a full 42RU. This allows you to scale from 3 up to 16 nodes with the right form factors. With the previous R1 rack version, it was only possible to run 3 to 5 nodes.
Ok, that sounds good. I can now scale without worrying about additional nodes (at least for the time being) and I have a standard design in the datacenter (42RU). What about the instance types?
There are several instances available. You can start small and go from 48 vCPU with 256GB of RAM up to 96 vCPU, 768GB RAM, and literally double the capacity of the NVMe storage (23TB). This last type of instance is the M1d.medium which has been announced in the release.
Note there is a current limitation, you cannot mix the type of instances (hopefully this might change in the near future).
The most dramatic change is the hardware dimension. Let’s go over the capacity planning and make the assumptions that:
- vCPU: physical core overcommit ratio is 2:1
- There is no memory overcommitment
- Storage offers 50% of the total capacity (RAID1)
The outcome is that we get 2x more VM density and 2x more workloads than previous versions:
Is there any difference with other offerings?
The biggest differentiator is how this new offering is packaged and priced. You have a pre-built, pre-installed, and pre-tested VxRail running VMware Cloud Foundation (VCF) solution. You don’t need to manage the hardware as it’s included in the subscription and you are billed on a monthly basis (you need to sign a commitment for a number of months).
There are a number of offerings in the market that can compete with this new solution, such as AWS Outpost, Google Anthos, Azure Stack, or Oracle Dedicated region Cloud@Customer. In some of them, the amount of resources is higher and you can also run VMware workloads, so the competition is huge.
If there are similar solutions, why should we pick this solution?
That’s a good question and it really depends on what you are looking for. There are many options that allow building hybrid architectures. Let’s take a look at a couple of alternatives:
What about VMware on AWS?
Both VMware Cloud on AWS and VMware on DELL EMC uses the same underlying VMware control plane and APIs. However, in case of hardware failure, how long does it take to replace one or the other solution? Remember that AWS uses its own custom hardware, so please be aware of that. In addition, Outpost relies on connectivity to the parent AWS region. If you have a WAN outage your services will keep running but you will not be able to manage, deploy, or monitor your infrastructure.
What about Oracle Dedicated Region?
If you want to have a full Cloud platform running behind your firewall, this is probably the best solution. Oracle provides the same architecture, services, and SLAs as you can find in their cloud offering. It is a fully-managed region that includes OCVS (Oracle’s VMware solution). The solution is based on VMware Cloud Foundation (VCF) and delivers the full software-defined center stack (SDDC). You have full control of the hypervisor as opposed to other cloud solutions. The only thing you need to consider is that this is a full cloud solution, meaning that you will need space in your datacenter to allocate all the racks.
As you can see, you now have a fully managed offering that comes pre-built, pre-installed (VMware VCF), and runs under a subscription model. This allows your IT teams to remove the painful task of managing the hardware and worrying about hardware refresh and upgrades. VMware will do the full lifecycle management.
You have other solutions in the market, which are good alternatives. However, it depends on your needs and what you are really looking at, this can be a good stack to deploy in your facilities (or co-located).