Understanding Cloud Costs Using Cost Analysis – Virtana

Virtana Migrate Services can provide you with a Cost Analysis of the estimated costs you would incur based on which Cloud Service Provider (CSP) you want to use. This allows you to model potential savings before you migrate your data to the cloud.

A sample of the output for a cloud cost comparison is available on the Cost Analysis page in Workload Placement. If you would like Virtana Migrate Services to create a similar cost analysis for your environment, contact a sales representative.

Cost analysis output is based on the resources in your move groups that you create in Workload Placement. In addition to selecting a CSP, you can set a few other parameters to refine the cost estimates, such as the region in which your resources would reside, the headroom allotted to VMs, etc. Virtana Platform keeps current with the continuous changes in CSP costs and offerings.

Analysis Process

On-premises usage is discovered and is converted to public cloud costs per the VMs in the move group you choose to analyze. Costs are based on the pricing policy of each cloud platform.

It is particularly important to understand your network and I/O usage because network and I/O incur significant costs in the cloud, as compared to on-premises.

Some of the factors considered in the analysis process include the following:

Physical

Parent host data
Hardware make and model
CPU type, cores, and speed
Memory
Storage capacity and storage type

Virtual (VMs)

Parent host data
vCPUs
Memory
Storage
Subnet
Guest OS
Power-on state

Utilization history

CPU
Memory
Network activity
Disk I/O activity

CSP services and features considered in the recommendation process:

Selecting optimal compute type
Selecting optimal license type
Right-sizing
Application uptime
Elastic architecture
Windows licensing
Bring Your Own License (BYOL)
Platform as a Service (PaaS)
Serverless Compute / Containers as a Service (CaaS)
VMware on AWS, Azure, GCP, and IBM Cloud

Cloud Mapping

During migration planning, the assets to be moved to the cloud are put into move groups. To determine the cost of moving your assets to the cloud, your current on-premises assets must be mapped into the cloud based on identified workloads. To determine workload, we leverage CPU data, Network data, Storage data, Host information, and VM information.

Virtana Migrate Services performs a vendor-agnostic cloud mapping and cost search in the CSPs' latest published specifications. They determine the rightsized cloud mapping or configuration (including compute, storage, and egress network traffic) at the lowest possible cost.

During calculations, the migration services team does the following:

Determines which type of compute device (CPU and ephemeral memory) can accommodate the workload.
References a specific CSP's price list and maps to a corresponding hardware type.
Uses the cloud provider’s price list to determine the corresponding monthly cost for this VM, based on the hardware type.

The storage hardware types used for each CSP are:

AWS: EBS General Purpose SSD (gp3) and Provisioned IOPS SSD
Azure: Managed Ultra Disk, Premium SSD, Standard SSD, Standard HDD

Custom Pricing for Cost Analysis

The Virtana Platform allows a user to specify a custom price list for any project, prior to importing the project files into Workload Placement. This price list is then used, instead of the standard defaults, to compute the on-premises cost for ALL move-groups within the project.

To add custom pricing to the Workload Placement calculations, you must download the Migration Analysis files from VirtualWisdom. The custom_price_list.csv file contained in the archive file must be modified to include cost details for your infrastructure. The entire archive file is then uploaded to Workload Placement as a new project.

For details, see Adding Custom Pricing for Migration in the VirtualWisdom User Guide.

In-cloud Costs

The following rules are applied in the calculation of in-cloud VM cost.

Compute costs are determined by first mapping the VM to a best-fit cloud instance, based on CPU and memory usage.
After a Cloud instance has been determined, Virtana Migrate Services then determines if the instance can support the required storage capacity. If not, it is remapped to a different instance that can support the required storage capacity.
VMS will find an equivalent managed disk for each disk attached to the virtual machine. For example, in VMware vDisks are attached to virtual machines. If a VM has 8 vDisks attached to it, then VMS will attempt to find the equivalent 8 AWS EBS or Azure managed disks that can meet the throughput, IOPS, and capacity for each disk.
VMS will report “Pass” and the type of disk and “fail” if it could not find a CSP’s disk that can match the original on-premises performance specifications.
Mapped storage is standard EBS or equivalent as described below:
- AWS: EBS General Purpose SSD (gp3)
- Azure: Managed Disk Standard SSD

On-premises Costs

The following rules are applied in the calculation of on-premises VM costs.

On-premises costs are determined from the VM perspective. VM compute cost is a relative cost allocation. The intent is to compare on-premises VM costs to in-cloud VM costs.
The on-premises cost of a VM is a fair allocation of the parent host-cost, based on the relative size of the VM with respect to its parent host. The following factors are considered as the costs are calculated:
- Lifetime of the host hardware :
  
  Over which the server hardware is amortized (default = 36 months). This is used to calculate the monthly cost.
- CapEx1:
  
  Total cost of the hardware including any applicable taxes, network, implementation costs, etc., minus the cost of storage.
- CapEx2(storage cost):
  
  Total cost of DAS or SAN (or iSCSI) storage or NAS devices if the VMDK files reside in a NAS device.
- OpEx:
  
  All operating costs such as facilities, power, WAN, backups, administration and other labor, etc.

For default calculations, on-premises host and VM costs are estimated based on a customer’s hardware configuration compared against industry benchmarks and norms. These costs are based on a default lifetime of 36 months. However, you can apply your actual costs to provide a more accurate view of your on-premises costs. See Custom Pricing for Cost Analysis for more information.
Calculating on-premises VM costs
- CapEx1:
  
  A portion of the total Capex1 is allocated to the VM based on the VMs relative size (CPU & Memory) to the parent host (CPU & Memory). It factors the following:
  - Excess capacity:
    
    For example, if a host has 24 cores and 1TB RAM but has only 1 VM with 2 CPU and 8GB RAM, then only the 2 CPUs and 8GB RAM will be considered in calculations.
  - Move Group aware:
    
    If your move group includes only a subset of the VMs in a host, then host cost will consider only the VMs in the move group.
  - Overcommit aware:
    
    It is very common for customers to over-allocate CPU and memory. For example, if a host has 24 cores and 1TB RAM and the CPU/memory has been overcommitted by a factor of 2, then the host is treated as a system with 48x2TB system, effectively reducing the VM costs by 50%.
- CapEx2 (storage cost):
  
  VM storage is an absolute cost assignment based on the allocated (not used) VM storage. For example, assuming you have a host with 2 VMs and the following configuration:
  - Host has 10TB storage costing $20,000.00 - effective cost $2/GB
  - If VM1 has been allocated 500GB, then the on-premises VM storage cost for this VM would be $1,000.00
  - If VM2 has been allocated 2TB, then its storage cost would be $4,000.00
  Note: Allocated storage is strictly from the point of view of the VM as storage seen within VMDK(s). This applies regardless of how the VMDK has been provisioned (in DAS, SAN, NAS). Network storage connected to a VM is not considered as the VMs native storage and therefore is not included in the mapping/costing.

OpEx:

It uses the same allocation rules as CapEx1
Lifetime:

After the cost of the host has been allocated to a VM, then the monthly VM cost is calculated using this value.

Network Usage Costs

Network usage costs charged by cloud service providers (CSPs) can be confusing. They can also be some of your most significant cloud costs. It is important to understand the cost of moving workloads or applications to the cloud while other applications remain on-premises. Virtana Migrate Services provides you with total cloud cost for each move group you are considering for migration to the cloud.

Network traffic that occurs between VMs within a move group (in the cloud) is not considered as egress traffic. Nor is traffic from on-premises to the cloud. Traffic from the cloud to on-premises does incur costs, as well as traffic that was originally existing on-premises to public IP endpoints.

CSPs charge for what they consider "egress" traffic. Egress traffic is outbound traffic that leaves the network boundaries of the cloud provider’s data center. Virtana Migrate Services provides the total outbound (egress) network cost of compute devices for a given move group.

The computed network cost reflected in the Workload Placement reports includes network cost leaving the data center only for the list of selected VMs (compute instances) within the move group.

The computed egress is updated for each change you make to a move group topology and is reflected in the new cost calculations for the specified move group.

This means that egress cost for any VM can differ, depending on the communication flow to and from the VM.

move groups