2.4 Performance

Performance for migrations using PlateSpin Migrate depends on many factors. Use the guidelines in this section to understand those factors and better plan your migration projects.

2.4.1 Performance Characteristics

The performance characteristics of your PlateSpin Migrate product depend on a number of factors, including:

  • Hardware and software profiles of your source and target

  • Hardware and software profiles of your PlateSpin Server host

  • Hardware and software profiles of your target virtualization host or cloud host environment as VMs compete for resources

  • The specifics of your network bandwidth, configuration, and conditions

  • The number of your source workloads’ volumes and their sizes

  • File density (number of files per unit of capacity) on your source workloads’ volumes

  • Source I/O levels (how busy your workloads are)

  • The number of concurrent migrations and the number and type of the targets

  • Whether data encryption is enabled or disabled

  • Whether data compression is enabled or disabled

For planning large-scale workload migrations, you should perform a test migration of an average workload and use the result as a benchmark, fine-tuning your metrics regularly throughout the project. In addition to the data transfer process, also consider the other phases that a migration job goes through, as applicable to your project:

  • Preparation and network setup

  • Source workload and target machine discovery

  • Target configuration

2.4.2 Scalability

You can set up multiple workload migrations and run them concurrently. See Performance Characteristics for information about the many factors that impact performance of PlateSpin Migrate in your migration environment.

Concurrent Replications and Migrations

Performance for concurrent replications and concurrent migrations depends on the resources on the PlateSpin Migrate server and the target environment, as well as the available bandwidth. We recommend that you begin with a low load, then increase it and see how the migrations perform in your environment. Use the scheduled start dates to control when migrations begin and how many migration jobs are scheduled to run concurrently.

The available hardware resources on your Migrate server impact the number of managed workloads and concurrent replications the server can handle. Generally, the higher the load is for concurrent replication and migration, the more resources it consumes.

Scalability testing performed with VMware ESX hosts suggests the following benchmark recommendations:

  • Multiple migrations to a single VMware ESX Host server: no more than 10

  • Multiple migrations against multiple VMware ESX Host servers: no more than 40

In a VMware Cluster, ensure that you balance migrations across multiple hosts in the cluster for best performance.

Workload Discovery and Inventory

We recommend that you keep no more than 50 discovered workloads at a time in the inventory for your PlateSpin Migrate server, depending on its available hardware resources. As you complete workload migrations, you can remove workloads and add others.

You cannot necessarily concurrently run replications and migrations for all the workloads in your inventory. Use the scheduled start dates to control when migrations begin and how many migration jobs are scheduled to run concurrently. See Concurrent Replications and Migrations.

PlateSpin Migrate provides three discovery tools:

  • Migrate Web Interface: Discover one workload at a time.

  • Migrate Client: Discover one workload at a time, multiple workloads at a time, or all workloads in a domain.

  • Mass Discover CLI: Discover one or multiple workloads from a CSV file.

For more information, see About Source Workload Discovery.

2.4.3 Data Compression

If necessary, PlateSpin Migrate can compress the workload data before transferring it over the network. This enables you to reduce the overall amount of data transferred during a workload migration job.

Compression ratios depend on the type of files on a source workload’s volumes, and might vary from approximately 0.9 (100MB of data compressed to 90 MB) to approximately 0.5 (100MB compressed to 50MB).

NOTE:Data compression utilizes the source workload’s processor power.

Data Compression can be configured per migration job. You can also use the PlateSpin Migrate Client to specify a default compression value to be applied globally. See Section 6.2, Configuring Job Values Defaults.

To set the data compression level for the migration job using the PlateSpin Migrate Web Interface, see Compression Level setting in the Section 25.1.1, Configuration Workflows Using Migrate Client.

2.4.4 Bandwidth Throttling

PlateSpin Migrate enables you to control the amount of available bandwidth consumed by direct source-to-target communication over the course of a workload migration. You can specify a throughput rate for each migration job. You can specify whether to throttle at all times or on specific days of the week and times of day. This provides a way to prevent migration traffic from congesting your production network and reduces the overall load of your PlateSpin Server.

Bandwidth throttling is a parameter of a workload migration job’s configuration properties. To apply bandwidth throttling for the migration job, see Bandwidth Throttling during Data Transfer.

2.4.5 Blackout Window

PlateSpin Migrate Web Interface enables you to specify a blackout window for replication. The blackout window suspends scheduled replications from starting during a specified period of time and pattern. It helps you to reserve network bandwidth for users or mission critical communications during peak traffic periods. You can also use it to prevent conflicts for other data backup or snapshot activities.