Block-Level Replication FAQ

  • 7920943
  • 24-May-2007
  • 01-Aug-2017

Environment

PlateSpin Protect 10.x and above
PlateSpin Forge 3.x and above
PlateSpin Migrate 9.x and above

Situation

This document provides information regarding PlateSpin Block-level replication which offers the ability to relocate and replicate servers without any service interruption to the source workload (once the block level filter driver has been installed).

Resolution

Block-level replication is designed to migrate and protect Windows workloads without requiring the source workload to be taken offline. Block-level replication is most suitable for servers with high transaction activity (for example, database transactions) as during the transfer, the block level mechanism will copy only the changed blocks instead of the entire file.

Supported Source Operating Systems for Block-Level Replication:

Windows 2003 Server (SP1+)
Windows 2008 Server
Windows 2008 Server R2
Windows 2012 Server
Windows 2012 Server R2

Supported Job Types:

Physical-to-Virtual (P2V)
Virtual-to-Virtual (V2V)
Physical-to-Physical (P2P)*
Virtual-to-Physical (V2P)

Note: Forge and Protect do not offer P2P functionality.

Technology Overview:

When configuring a job to use Block-level replication, the following will occur:

1. A block level filter driver will be installed on the source (source workload will reboot once). This filter driver will run in Kernel mode and will be able to copy locked files, used files, etc.  Note: if using Migrate, or if the source workload is a cluster, the filter driver is not Kernel mode and no reboot is required.
2. The filter driver will generate a bitmap view of the source volumes, assigning 1 bit for every block (the .bbvt file).
3. During the transfer to the target, the filter driver will detect any changes to the volume and the changed blocks will be copied again until full synchronization is achieved between the source and the target.
4. If using Migrate the filter driver will be removed from the source (no reboot required).  If using Forge or Protect, the driver is left in place for future replications.  Note: removing the Kernel mode driver, either by opting to when removing the workload or manually, will require a reboot of the source workload.

Frequently Asked Questions:

Which Target Servers can I use when configuring a job with Block-level replication?

Any supported target can be used for Block-level migrations/replications, with one exception: Block-level is not supported when capturing or deploying an Image.

Can I relocate and protect workloads with Dynamic Disks?
Yes. Block-level replication offers full support for Dynamic Disks (including Mirroring, Spanned Volumes, RAID5, etc.). Note: the target workload will have a Basic disk only.

What are the best scenarios to use Block-level replication?

Block-Level replication is recommended for the following scenarios:

Windows Workloads with high disk, application and database activity – Block-level replication is recommended for these servers as it copies only the actual blocks which were changed during the transfer step instead of the entire file. For example, if a 1 GB database file was changed during copy, the block level filter driver will mark these blocks and will copy only these blocks (File level transfer would have copied the entire 1 GB file again). This process will significantly reduce the network traffic and the total migration time.

Windows workloads located in a network with high latency and low bandwidth – Block-level replication is recommended for networks with high latency and low bandwidth as it copies less data between synchronization attempts. Furthermore, the transfer protocol is able to produce high transfer rates when compared to other transfer protocols (e.g. Take Control transfer and Live file based transfer protocol).

Workload Migration or Workload Protection jobs where service downtime is not possible – Since Block-level replication is able to copy locked and in-use files, source workload services will not be stopped during relocation or incremental backup jobs.

What job features are not available when using Block-level replication?

Disk Resizing – if Block-level replication is selected, Disk resizing will be disabled, although selecting which volume to copy is available

Can I use Block-level replication to create PlateSpin Flexible Image?

Block-Level Migration is available only for Windows workload Peer-to-Peer jobs. PlateSpin Flexible Images can be created with Take Control or Live file based transfer options.

Can I transfer SQL, Exchange, Oracle, Sybase or other proprietary applications with Block-level replication?

There are no restrictions to the types of applications for Block-level replication. Block level transfer will copy the entire volume to the target and will provide a ‘crash consistent’ state volume on the target.

What are the best practices for Block-level replication jobs?
Block level transfer relies on several technologies and protocols. To ensure the source workload and the environments are compatible with block level replication, PlateSpin recommends following the below before running a job using block level replication:
 
1. Ensure the source workload operating system is compatible with Block Level Transfer as per the list mentioned above
2. Ensure the RPC Service is running on the source and that the RPC Protocol between the source and the target servers is not blocked by Firewall.
3. Check with your application vendor if information is cached in the memory before performing a workload relocation job. If needed, shutdown the application and the related services to ensure the highest level of synchronization