Estimating vCenter Server 5.5 to vCenter Server Appliance 6.0 migration time
search cancel

Estimating vCenter Server 5.5 to vCenter Server Appliance 6.0 migration time

book

Article ID: 341774

calendar_today

Updated On:

Products

VMware vCenter Server

Issue/Introduction

vSphere 6.0 Update 2m enables you to migrate a source Windows vCenter Server 5.5 deployment to a target vCenter Server Appliance 6.0 Update 2m with an embedded PostgreSQL database. The source Windows vCenter Server 5.5 database, either embedded or external, will be migrated to PostgreSQL. The migration time depends on data size, data retention policy, hardware, and other factors including network and storage.


Environment

VMware vCenter Server 5.5.x
VMware vCenter Server Appliance 6.0.x

Resolution

This article provides information for estimating the downtime required for the vCenter Server to vCenter Server Appliance migration. For more details on the vCenter Server migration process, see vSphere 6.0 Documentation Center.

Database size estimation

Estimating the migration time for the vCenter Server database depends on these factors:
  • The database size
  • Host hardware and configuration
  • Whether performance and historical data is migrated
To determine the right options for the environment, VMware recommends that you examine the size of the database in order to estimate the migration time. To assist in estimating the size of the database, use the attached SQL 2146420_QueryDBSize_MSSQL.sql or 2146420_QueryDBSize_ORACLE.sql query to gather the data size of inventory and configuration data and stats, events, alarms, tasks (SEAT) data.

Impact of database size on vCenter migration time

Use the table below to estimate the vCenter Server migration time for the database. The data is based on lab test results (with the lab configuration in Appendix 2). With different hardware configurations, migration time may vary.

Database size(GB)

40

75

150

250

Migrating Inventory and Configuration data only (7GB database)

30 minutes

30 minutes

30 minutes

30 minutes

All Data Migrated (including performance and other historical data)

2 hours 46 minutes

5 hours 28 minutes

11 hours 45 minutes

20 hours 47minutes


In the table above, for DB size value, the 7GB inventory and configuration data is based on vSphere 5.5 configuration maximums of 1000 hosts and 10,000 powered-on Virtual Machines. For more information on configuration maximums, see the vSphere 5.5 Configuration Maximums guide. The remaining data is divided equally between Event/Task data and performance statistics data.

For example:

With 40GB data size in the table, 7GB is inventory data, 16.5GB is Event/Task data and the remaining 16.5GB is stats data.

To assist in time estimation, see the attached file name 2146420_TimeEstimation.xlsx, which provides a time estimation formula.

As an alternative to using the attached 2146420_QueryDBSize_MSSQL.sql or 2146420_QueryDBSize_ORACLE.sql query and 2146420_TimeEstimation.xlsx time estimation tool separately, use the attached 2146420_Get-VCDBUsage.ps1 PowerShell script. This script integrates the database size query and migration time estimation in a single script to simplify the time estimation process.

Notes:
  • Running locally on the Microsoft SQL Server database requires Windows PowerShell Extensions for SQL Server. If the Invoke-Sqlcmd does not exists on the system, an error will be thrown and the script will exit. To correct this, either install Microsoft SQL Tools or manually run the SQL query on your VCDB.

  • Running remotely connected to the Oracle database requires the Oracle ODAC Client to be installed on the Windows system running the script.
The script will extract the current database usage and calculate the estimated amount of time needed to migrate the VCDB as part of the Windows vCenter Server to vCenter Server Appliance migration. Both options and their estimated times will be provided at the end of the script.

Impact of hardware configuration on vCenter migration time

For the database export phase:
  • In 6.0 U2m, the source vCenter Server CPU speed may impact export time due to the migration process utilizing only one of the CPUs in the source vCenter Server machine, and usage for that CPU may be high.
    • When using a slower CPU compared with the lab test in Appendix 2 for the Source vCenter Server machine the exporting time could be longer than estimated.

  • Source vCenter Server database I/O is another major factor that affects export time.
    • If there are disks with lower maximum achievable IOPS than the lab configuration on the source vCenter Server database machine, expect that the exporting time could be longer.
For the import phase:
  • The importing phase includes CPU-intensive operations, and both CPU count and CPU frequency in the destination vCenter Server Appliance will impact importing time.
    • If there are slower CPU compared with the lab, or have fewer CPUs for the destination vCenter Server Appliance, expect that the import time could be longer.

  • The destination vCenter Server Appliance I/O can affect import time.
    • If there are disks with lower maximum achievable IOPSs than our lab configuration on the destination vCenter Server Appliance machine, we expect that importing time could be longer.

  • VMware recommends using a 1Gbps or higher speed network to perform the vCenter Server migration.
  • Ensure the hardware configuration (CPUs, memory, etc.) meet the minimum requirements for a vCenter server setup. For more information, see:

Using these variables and information provided in this article, create an estimation of the time required to migrate the vCenter Server.

Migration options

Based on this information, these migration option are available:
  1. Migrate inventory and configuration data only.

    Note: In vSphere 6.0 Update 2m, if you choose to migrate inventory and configuration data only, historical data (event/task/stats) cannot be migrated later.

  2. Migrate all the data, including performance and other historical data.

    Note: This can significantly increase the maintenance window required.
If the estimated migration time is too long, truncate the SEAT data to a smaller size to reduce the maintenance window. For more information, see Purging old data from the database used by vCenter Server (1025914).

APPENDIX 1: SQL queries to get the size of core and SEAT tables

You can use the attached SQL query (2146420_QueryDBSize_MSSQL.sql or 2146420_QueryDBSize_ORACLE.sql) to gather the data size of core and SEAT tables. The following screenshot is an example output from the SQL queries to get the size of core and SEAT tables.

Notes:
  • Migrating the inventory and configuration only, then core and alarm data in the table below will be migrated.
  • Migrating the performance and other historical data as well, then all the data in below table will be migrated.

APPENDIX 2: Performance test and results

VMware conducted lab testing to measure the performance of the vCenter Server migration process.

Test environment and configuration:
  • The test environment contains three ESXi servers:
  • The first server hosts the target vCenter Server Appliance (VCSA).
  • The second server hosts the source Windows vCenter Server virtual machine.
  • The third server hosts the source vCenter Server's external SQL database.
The figure above illustrates the experimental testbed along with the migration data transfer flows.

The following list summarizes the hardware and software specifications of the testbed.
  • Physical ESXi host1 (ESX1), ESXi host2 (ESX2), and ESXi host3 (ESX3)
    • Dell PowerEdge R730xd
    • CPUs: Two 12-core Intel Xeon E5-2680 v3 @2.50GHz, Hyper-Threading enabled
    • Memory: 256GB
    • Network Card: Intel X540 10Gb BT Network Card
    • Storage: 6 x 400GB Intel s3710 SATA 6Gb/s 2.5in SSD
    • Virtualization Platform: VMware vSphere 6.0.0 GA (Build #2494585)

  • ESX1 hosts the Target vCenter appliance virtual machine:
    • Virtual machine configuration: 16 vCPU, 32GB of memory
    • vCenter Server version: 6.0 U2m
    • Embedded PostgreSQL vCenter database
  • ESX2 hosts the Source vCenter Server virtual machine:
    • Virtual Machine configuration: 16 vCPU, 32GB of memory
    • Guest OS: Windows Server 2012 R2 Standard
    • vCenter Server version: 5.5 U2 GA (Build #2105955)
    • Inventory size: 1K hosts, 15K VMs
    • Running migration assistant application (MA)
  • ESX3 hosts the External source vCenter database virtual machine:
    • Virtual Machine configuration: 16 vCPU, 32GB of memory
    • Guest OS: Windows Server 2012 R2 Standard
    • Microsoft SQL Server 2012
    • Data sizes in source vCenter Server database virtual machine
      • DatasetSize-1 includes
        • 7 GB inventory and configuratione data
        • 35 GB task and events historical data
        • 35 GB statistics historical data
      • DatasetSize-2 includes
        • 7 GB inventory and configuratione data
        • 70 GB task and events historical data
        • 70 GB statistics historical data
Lab Test Results for vCenter Server Migration

Below we list is the performance number for migration tests in the above environment,

For DatasetSize-1, total end-to-end time is 5 hours 28 minutes.
  • Exporting phase: 1 hours 32 minutes
  • Importing phase: 3 hours 56 minutes
For DatasetSize-2, total end-to-end time is 11 hours 40 minutes.
  • Exporting phase: 2 hours 53 minutes
  • Importing phase: 8 hours 47 minutes
VMware also collected resource usage data for the migration tests. By default, we present the average value, and we also list some maximum usage data for reference.


CPU Used (%) Across All CPUs

Memory Used (GB)

Disk read and write/s (IOPS)

Network Throughput (Mbps)

Source vCenter Server

7

17

average 75, max 9800

average 56, max 1692

Source vCenter Server Database

0.7

16

average 271, max 5487

average 56, max 1688

Target vCenter Server Appliance

9

5

average 3247, max 87160

average 8, max 1016

Attachments

12146420_TimeEstimation.xlsx get_app
2146420_QueryDBSize_ORACLE.sql get_app
2146420_QueryDBSize_MSSQL.sql get_app
2146420_Get-VCDBUsage.ps1 get_app