The talk of VMworld 2017 in Las Vegas was the ability to run VMware Cloud on AWS. But how does the cost of running...
VMware Cloud on AWS compare to simply running EC2 instances directly on Amazon Web Services?
VMware Cloud on AWS pricing
VMware Cloud on AWS pricing is based on two main factors -- the number of hosts provisioned and whether you opt to use on-demand or reserved consumption.
On-demand consumption is exactly what it sounds like. You can provision a VMware infrastructure on an as-needed basis and pay by the hour for the resources you use. Conversely, reserved pricing is billed at a lower rate, but requires you to commit to using the VMware resources for either one year or three years.
As was previously noted, VMware Cloud on AWS pricing is based on the number of hosts you use. Each host includes two CPUs with 36 cores -- 72 hyper-threads. The hosts are also equipped with 512 GB of RAM, 3.6 TB of nonvolatile memory express (NVMe)-based cache and 10.7 TB of raw NVMe storage.
Because VMware Cloud on AWS is based on the use of clusters, you won't be able to perform a single-host deployment. A cluster requires at least four hosts, but you can add hosts to the cluster in single-host increments.
The cost per host can vary by region. The effective hourly rate in the U.S. West region is as follows:
- On demand: $8.3681 per hour
- 1 year reserve: $5.9346 per hour
- 3 year reserve: $4.1616 per hour
Hence, the hourly cost decreases when you pay for a reserved cluster. Keep in mind that these rates are per host, not per cluster. If you need some help determining the overall cost, VMware provides a calculator.
AWS EC2 pricing
Amazon bills its Amazon Web Services (AWS) Elastic Compute Cloud (EC2) customers based on the resources consumed. Like VMware Cloud on AWS pricing, EC2 users can opt for on-demand or create reserved instances -- there is even an option to use dedicated hosts.
EC2 pricing tends to be much more complex than VMware Cloud on AWS pricing. The cost of an EC2 instance is based on factors such as the instance size, OS, region, and whether the instance is running on demand or reserved. To give you a general idea of the costs, a Windows Server instance -- t2.xlarge -- with 16 GB of RAM and four virtual CPUs running in the U.S. East region costs $0.2266 per hour. A comparable Linux instance costs $0.1856 per hour.
Obviously, the per hour cost is far lower than that of running VMware Cloud on AWS, but remember that this is the cost of a single VM instance, not an entire host server. A host server can run multiple instances.
So what if you were to go with an AWS reserved host? Based on the Dedicated Host Pricing sheet, a 36-core dedicated host (G3) in the East Region costs $5,016 per hour when billed on demand. As is the case with VMware, you can opt to reserve the host for one year or three years. The price varies based on the length of the reservation and whether or not you pay upfront. The lowest cost for a G3 host -- paid upfront for three years -- is $2.225 per hour.
One thing to keep in mind is that the pricing sheet doesn't list the full hardware specs for AWS dedicated hosts, so there could be hardware differences between AWS hosts and VMware hosts.
Which is less expensive?
If you only consider hourly costs, then VMware Cloud on AWS is more expensive than using EC2 instances. This is especially true for organizations that only need a small number of EC2 instances.
Although it's tempting to look at the raw numbers alone when determining the cost of running workloads in the cloud, there are often non-tangible costs that should also be considered. For example, if your entire on-premises infrastructure is based on VMware, then the efficiency of being able to operate seamlessly with VMware Cloud on AWS using the same management tools you already use every day might completely justify any difference in cost.