The topic of VMware cloud solutions comes up often in my work for two main reasons:


    • Cloud vendor comparisons around features and pricing. This includes minimum/maximum configuration specs, corresponding pricing, and cost-per-resource, and an illustrative high-level comparison across the major VMware cloud solutions. This will be covered in my article below.
    • Oracle license & compliance management in various cloud-based VMware solutions. This topic is largely covered by my previous posts on Oracle licensing in third-party cloud environments. However, I will revisit this specifically in the VMware cloud solution context in a subsequent article – so stay tuned.

Some History.
Since its founding in 1998, VMware’s virtualization technologies have become ubiquitous in the enterprise IT landscape. Before the arrival and rapid growth of cloud solutions, VMware had established itself as the incumbent infrastructure technology that allowed for speed, agility, resource optimization and operational standardization.

The arrival of the cloud, thanks to AWS, posed a challenge for organizations already wedded to the VMware stack. Customers wanted the flexibility and agility of AWS while retaining their VMware “way of doing things”. VMware itself made a brief but unsuccessful foray into the cloud with its vCloud Air offering circa 2013-2017. It didn’t go well.

In 2017, AWS and VMware joined forces to offer the VMware on AWS cloud solution. This service provided an AWS-based infrastructure where customers could build their VMware SDDC and bring their existing VMware-based workload seamlessly into AWS and into the cloud age. Since then, VMware has partnered with several leading cloud providers to provide similar “VMware in cloud” solutions. These include Microsoft Azure, Google Cloud Platform (GCP), and Oracle Cloud Infrastructure (OCI). 

In this article, we will take a high-level look at these services, their basic configurations, and pricing comparisons. We will be limiting the discussion to the following clouds: AWS, Azure, GCP, and OCI. I want to clarify at the outset that this is not a “best of” assessment. It is also not a “cheapest of” assessment. There are too many factors beyond the scope of this article to consider in order to identify the “best” or “cheapest” option. I hope this information will help any reader navigate the basics of VMware-in-cloud offerings and make informed decisions around pricing and key terms and considerations.  

General Concept.
Summarized for simplicity’s sake, the VMware cloud solutions generally consist of at least 2-3 physical nodes clustered together to form a VMware cluster. Each node has a fixed number of physical cores, memory and storage (typically high-performance SSD and/or NVMe). Scaling up requires adding more nodes to the configuration up to a maximum number of nodes defined by the cloud service offering. This maximum varies from vendor to vendor. If a VMware cluster is maxed out on its nodes, you can simply create additional clusters with the cloud vendor. Pricing is typically on a per-node (AWS, Azure, GCP) or per-core (OCI) basis. 

Also note that apart from GCP, all other vendors provide various node/host server specifications. AWS and Azure offers two types, while OCI offers 3. For each vendor, additional capacity must be added in the form of additional nodes of the same type.

Let’s start with a look at the configuration options across the vendors:

Vendor VMware Service Name Minimum # of Nodes per Cluster Maximum # of Nodes per Cluster Node/Host Options
AWS VMware Cloud on AWS 2 16 i3.metal:  Intel Xeon E5-2686 v4 (2.3Ghz): 36 cores per host / 512 GB RAM / 10.7 TB NVMe storage
i3en.metal: 3.1 GHz all core turbo 1st or 2nd generation Intel® Xeon® Scalable (Skylake or Cascade Lake) processors: 48 cores per host / 768 GB RAM / 45 TB NVMe storage
Azure Azure VMware Solution 3 16 AV36: Dual Intel Xeon 18 core 2.3 GHz: 36 cores per host / 576 GB RAM / 3.2 TB NVMe Cache / 15.20 TB SSD storage
AV36P: Dual Intel Xeon 18 core 2.6 GHz / 3.9 GHz: 36 cores per host / 768 GB RAM / 1.5 TB (Intel Optane Cache) / 19.20 TB NVMe storage
AV52: Dual Intel Xeon 26 core 2.7 GHz / 4.0 GHz: 52 cores per host / 1536 GB / 1.5 TB (Intel Optane Cache) / 38.40 TB NVMe storage
GCP Google Cloud VMware Engine 3 32 ve1-standard-72: Intel Xeon 2.6 GHz (3.9 GHz Turbo): 36 cores per host / 768 GB RAM / 3.2 TB NVMe Cache / 19.2 TB NVMe storage
OCI Oracle Cloud VMware Solution 3 64 AMD Processor: AMD EPYC 7J13. Base frequency 2.55 GHz, max boost frequency 3.5 GHz.
Two-socket BM.DenselO.E4.128 with two CPUs each running 16 cores: 32 cores per host / 2TB RAM / 54TB NVMe storage
Two-socket BM.DenselO.E4.128 with two CPUs each running 32 cores: 64 cores per host / 2TB RAM / 54TB NVMe storage
Two-socket BM.DenselO.E4.128 with two CPUs each running 64 cores: 128 cores per host / 2TB RAM / 54TB NVMe storageNote: Oracle recommends VMware SDDCs deployed across availability domains within a region do not exceed a maximum of 16 ESXi hosts.

 A cursory read of the table indicates a likely lower-cost entry point with AWS (with it’s 2-node minimum) and more heavy duty capabilities brought by OCI. OCI’s maximum core and node counts, with high memory and storage should lend itself better for large VMware deployments.

Another observation I had was that OCI’s VMware service is available in all of its 30+ global regions. Likewise, Azure was available in a majority of its regions (I noted 26). AWS offered its VMware service in 18 regions, with GCP offering it in 13. They key point is that in the overall assessment of a cloud vendor, it is important to ensure that all required services are offered in the desired regions. For example, an Azure customer in Switzerland may be disappointed to learn that the Azure VMware service in not offered in either of Azure’s Swiss-based regions.

Another valid topic I’m not covering here relates to the way the VMware solutions are configured and delivered, and the level of control the customer has in each vendor’s offering.

Minimum Configuration Comparison.
When we configure minimum VMware configurations in each vendor, we get the following:

Vendor # of Nodes Total # of Cores Total Memory (TB) Total Storage (TB)
AWS US East (N. Virginia) 2 72 1 21.4
Azure (East US) 3 108 1.7 45.6
GCP (us-east4) 3 108 2.3 57.6
OCI 3 96 6 162

An immediate observation here is how much more memory and storage OCI provides by default. Depending on the workload type, if it’s memory and/or storage intensive, OCI has an obvious edge.

Minimum Configuration Pricing Comparison – On-Demand Pricing.
Next, let’s apply On-Demand list pricing (excluding any promotional rates):

Vendor Hourly Rate Total Hourly Cost Total Monthly Cost Total Yearly Cost Yearly Cost per CPU Core Yearly Cost per TB of Memory Yearly Cost per TB of Storage
AWS US East (N. Virginia) $8.36 $16.72   $12,205.60   $146,467.20   $2,034.27   $146,467.20  $6,844.26
Azure (East US) $9.21 $27.63  $20,169.90  $242,038.80  $2,241.10  $143,430.40  $5,307.87
GCP (us-east4) $9.29 $27.87  $20,345.10  $244,141.20  $2,260.57  $108,507.20  $4,238.56
OCI $0.30 $29.25  $21,762.89  $261,154.71  $2,720.36  $43,525.79   $1,612.07 

Note 1: OCI’s pricing is in $ per OCPU per hour. Other vendors are priced per node per hour.
Note 2: AWS, Azure and GCP define 1 month of usage as 730 hours. OCI defines 1 month of usage as 744 hours.

On a simple cost basis, AWS clearly has the advantage – by a wide margin. By offering a minimum starting point with 2 nodes, the overall cost is minimized. The per-core cost is also the lowest. Azure and GCP come in next, with OCI being the most expensive on a per-core basis. However, the economics is very different when we consider cost on a per-TB of memory and per-TB of storage basis. OCI is by far the most economical.

Minimum Configuration Pricing Comparison – 3-Year Pricing.
Now, let’s consider the other end of the pricing spectrum – 3-year pricing with all upfront where available. Unsurprisingly, AWS keeps the edge. While prices drop ~50% across the board, the order is unchanged. OCI’s advantage on memory and storage is unchanged.

Vendor Hourly Rate Total Hourly Cost Total Monthly Cost Total Yearly Cost Yearly Cost per CPU Core Yearly Cost per TB of Memory Yearly Cost per TB of Storage
AWS US East (N. Virginia) $4.15 $8.30   $6,059.00   $72,708.00   $1,009.83   $72,708.00  $3,397.57
Azure (East US) $4.30 $12.90  $9,417.00  $113,004.00  $1,046.33  $66,965.33  $2,478.16
GCP (us-east4) $4.65 $13.95  $10,183.50  $122,202.00  $1,131.50  $54,312.00  $2,121.56
OCI $0.17 $15.84  $11,784.96  $141,419.52  $1,473.12  $23,569.92   $872.96 


At this point, it is worth mentioning that OCI still has a response on the per-core pricing. Unlike the other vendors, OCI’s VMware offerings with larger core counts include a substantial discount in the hourly rate. When we configure the minimum number of nodes with 128 cores each, OCI narrowly beats out AWS on the per-core pricing, while conceding its price advantage in memory and storage to GCP and dropping to second in both of those categories, just ahead of Azure.

Vendor Hourly Rate Total Hourly Cost Total Monthly Cost Total Yearly Cost Yearly Cost per CPU Core Yearly Cost per TB of Memory Yearly Cost per TB of Storage

(3*128 cores per node / 6TB RAM / 162 TB storage))

$0.1056 $40.55  $30,169.50  $362,033.97  $942.80   $60,339.00   $2,234.78 


While AWS generally maintains the edge in per-core pricing and OCI does so in memory and storage, it’s worth noting that Azure and GCP are well within the mix and competitively priced. Besides, there are many other factors to consider when comparing VMware cloud solutions like CPU types, storage and memory performance, overall cloud ecosystem, ingress/egress costs, vnc/vpc configuration requirements, VMware versions and tools available, and on and on. In short, selecting a cloud vendor when VMware is involved requires careful and methodical assessment, along with mapping simulations to each vendor to accurately calculate the VMware future and its place in the overall cloud picture.

If you are comparing different cloud vendors for a first cloud contract, or are already making a move and would like optimize the mapping to cloud, or would like to reduce your ongoing cloud spend with detailed technical analyses and assessments, feel free to reach out for a chat.

Useful links: