Migration preferences for servers

This page describes the available migration preferences for servers in Migration Center.

Google Cloud options

Specify the target product to which you would like to migrate your infrastructure. You can specify either No preference, or one of the following products:

  • Compute Engine
  • Google Cloud VMware Engine
  • Sole-tenant nodes

If you select No preference, Migration Center provides you with recommendations based on the internal default preference logic, which might be updated periodically.

The target product you select and the preferences you set are used to generate the TCO report. Migration Center also lets you customize the preferences for the other non-selected products, which are used to generate the license report.

Google Compute Engine

If you select Compute Engine as your target product, specify the different machine series, disk types, and licensing options that you want to apply in your TCO reports.

  • From the Machine series list, select all the Compute Engine machine series that you want to include as possible targets for migration of your virtual machine (VM) instances. To choose the most suitable machine series for your business needs, review the machine families comparison table.

  • From the Disk type list, select your preferred disk type.

  • From the Microsoft SQL Server licensing list, select the license type for the SQL Server database deployments that you plan to host on Compute Engine.

Google VMware Engine

If you select VMware Engine as your target product, specify the requirements for the group of virtual machines allocated across VMware Engine nodes.

Machine series family

Select the machine series family that you want to use. If a machine series family is not available in the region you selected, a warning appears.

Storage only nodes

If you have storage-bound workloads, selecting Use when available adds additional vSAN storage to storage only nodes, instead of HCI nodes. This might help reduce costs in the TCO report, and should be selected in the majority of cases. If you have specific requirements, however, you can exclude storage only nodes from your target environment by selecting Don't use.

If storage only nodes are not available for the machine series you selected, this preference is ignored, and only HCI nodes are considered.

GCVE service type

Choose how you plan to manage the license by selecting one of the available options:

  • Fully licensed: The fully integrated VMware Engine service, which includes all of the hardware, software, support, lifecycle management (such as patching and break fixes), as well as physical hosting environment (power and cooling, rack/stack) for the full VMware private cloud environment. This doesn't entitle the customer to a perpetual ownership of the VMware Cloud Foundation software, but rather, it is sold as a service, similar to other Google Cloud products.

  • Portable License: Broadcom and Google Cloud support license portability of VMware Cloud Foundation to VMware Engine. In this model, customers can purchase subscriptions of the new VMware Cloud Foundation software from Broadcom and Broadcom partners, and flexibly use those subscriptions on VMware Engine, as well as on their own on-premises data centers. The remainder of the service (hardware, hosting environment, support, lifecycle management) is purchased directly from Google Cloud or Google Cloud partners.

GCVE Protected

Select if you want to use VMware Engine Protected to enable backup and disaster recovery on your VMware Engine nodes.

VMware Engine Protected increases the costs in your TCO report and might not be available in all regions. For more information, see Backup and DR Service pricing.

Compute

You can configure the level of overcommitment for both CPU and memory. Overcommitment lets you allocate more memory and CPU cores on your VMs than what is available in the physical nodes.

  • The CPU core overcommit ratio lets you specify an overcommit ratio between 1.0 and 8.0, with 0.1 increment. The default value is 4.

  • The Memory overcommit ratio lets you specify the following values: 1.0, 1.25, 1.5, 1.75 and 2.0. The default value is 1.

Deduplication and compression

If you use deduplication and compression in your VMware environment, the Storage savings ratio lets you specify the amount of storage you're expecting to save when you migrate to Google Cloud, based on your current values on-premises. The default value is 1.7.

Sole-tenant nodes

If you select sole-tenant nodes as your target product, specify the node types that you want to see as recommendations in your TCO reports.

Learn more about sole-tenant nodes.

Overcommit factor

If you want to configure overcommit in your target sole-tenant nodes, in the Overcommit factor section, specify a value between 1.0 and 2.0, with 0.01 increment.

Learn more about overcommitting CPUs on sole-tenant nodes.

Host maintenance policy

Specify the host maintenance policy that you want to apply in your target sole-tenant nodes. This might affect the required size of the sole-tenant node cluster that you see in the TCO report.

Learn more about host maintenance policy.

Sizing method

Select the sizing strategy that you want to apply to your VMs in the TCO report. Choose one of the following:

  • No preference: let Migration Center choose the sizing based on the system defaults.
  • No rightsizing: choose a target VM size that closely match the size of your current machines. The algorithm doesn't apply rightsizing, which means that it doesn't consider the current performance data of your machines.
  • Moderate: calculate your target VM size based on the current performance of your machines, and apply moderate rightsizing to your VMs. This approach balances between performance and cost.
  • Aggressive: calculate your target VM size based on the current performance of your machines, and apply aggressive rightsizing to your VMs. For example, if you underutilize your machine CPU, your TCO report will suggest a VM with fewer CPUs. This approach maximizes the use of the VM's resources and reduces costs, but might also impact performance in the event of an unexpected load.
  • Custom: manually configure the parameters that you want the algorithm to use for the calculations. When you choose the custom sizing optimization, you can change the following parameters:
    • Metrics baseline: the baseline for calculating memory, CPU, and disk IOPS.
    • Storage multiplier: the amount of storage that you want to provision in your infrastructure, as a proportion of the current storage utilization.
    • CPU usage: the CPU utilization that you want to have in your infrastructure, as a percentage of the current utilization.
    • Memory usage: the memory utilization that you want to have in your infrastructure, as a percentage of the current utilization.

The following table shows the calculation parameters for the Moderate and Aggressive optimization strategies.

Metric Moderate Aggressive
Metrics baseline 95th percentile 95th percentile
CPU usage 70% 90%
Memory usage 85% 100%
Storage multiplier 1.25x 1.10x
Uptime 730 hours per month 730 hours per month

Network

Specify the outbound traffic that you expect to have. Outbound traffic is the percentage of the total network traffic that leaves the Google Cloud network. You are only charged for outbound traffic.

Product pricing track

Specify your preferred pricing track for the VMs that you migrate to Google Cloud.

Choose between on demand pricing, resource-based committed use of one or three years, or flexible committed used of one or three years. Committed use discounts (CUDs) apply when you choose a committed use, and appear in your TCO report.

OS license type and pricing track

If you use Windows servers or specific Linux distributions, you can set your preferences for licensing and pricing track in a granular way.

Windows servers

If you select sole-tenant nodes as your target product, you can select one of the following license types for your Windows servers:

  • Pay as you go
  • Bring your own license
  • No preference

If you select Compute Engine as your target product, only the Pay as you go (PAYG) option is available.

Optimize your license costs

By default, Compute Engine runs two vCPUs on each physical CPU. This means that for each physical CPU in your target environment running a Windows workload, you pay for two Windows licenses, which are applied per vCPU.

To reduce the costs of your target servers, you can choose to disable simultaneous multithreading (SMT). Migration Center lets you specify the following options for SMT:

  • Auto: Let Migration Center choose the cheapest among the available options. This is the default.
  • Enable: Leave SMT enabled as per default, and calculate the costs of two licenses per physical CPU.
  • Disable: Disable SMT on the physical CPUs to save on the license costs. This might impact the performance of your servers.
  • Disable with compensation: Disable SMT on the physical CPUs, but compensate for the missing vCPUs, for example by increasing the target number of CPUs, or by switching to a machine family with higher performance.
  • No preference: Use Migration Center default preference algorithm.

When you apply this preference, consider the following:

  • Your SMT preference is applied on top of rightsizing. Therefore, if you select Disabled, Disabled with compensation or Auto, this might increase the target utilization above the target utilization that you set for rightsizing.
  • If there is no target VM series that is compatible with the selected SMT preferences, then the SMT preference is ignored.

For more information about SMT in Compute Engine, see Set the number of threads per core.

Linux servers

Migration Center lets you specify the license type and pricing track for the following Linux distributions:

  • Red Hat Enterprise Linux (RHEL)
  • SUSE Linux Enterprise Server (SLES)
  • SUSE Linux Enterprise Server (SLES) for SAP

For the pricing track, you can choose between on demand pricing, one-year committed use, or three-year committed use.

What's next