QoreStor 6.1.0 - User Guide

Enabling MultiConnect

Before using MultiConnect, ensure that the default port configuration is appropriate for your environment. The port used by MultiConnect is:

• 11000 - This is the standard MultiConnect communication port for backup.
• 9920 - This is the standard MultiConnect communication port for managed replication.

To enable Secure Connect on a Windows client

1. On the client server, press Win+R to open the Run window.
2. Type sysdm.cpl and click OK.
3. Click the Advanced tab, then Environment Variables.
4. In the System Variables section, click New.
5. In the Variable name field, enter REMOTE_CLNT_MAX_CONNS.
6. In the Variable value field, enter one of the following:
   • 4 - establishes 4 connections.
   • 8 - establishes 8 connections.
   • 16 - establishes 16 connections.
7. Click OK, then OK.
8. Restart the DMA services for the change to take effect.

To enable MultiConnect on a Linux client

1. At the command prompt on the client machine, enter the following command

   echo 'export REMOTE_CLNT_MAX_CONNS=<4|8|16>' >> /etc/profile

   Where:

   • 4 - establishes 4 connections.
   • 8 - establishes 8 connections.
   • 16 - establishes 16 connections.
2. Restart the DMA services for the change to take effect.

To enable MultiConnect between QoreStor servers

1. At the command prompt on the source machine, enter the following command

   echo 'export REPL_CLNT_MAX_CONNS=16' >> /etc/oca/oca.cfg

2. Restart the ocards service for the change to take effect.

Configuring and using Rapid NFS and Rapid CIFS

Rapid NFS and Rapid CIFS enable write operation acceleration on clients that use NFS and CIFS file system protocols. Similar to OST and RDS, these accelerators allow for better coordination and integration between QoreStor backup, restore, and optimized duplication operations with Data Management Applications (DMAs) such as CommVault, EMC Networker, and Tivoli Storage Manager. For the current list of supported DMAs, see the QoreStor Interoperability Guide.

Rapid NFS is a new client file system type that ensures that only unique data is written to QoreStor. It uses user space components and file system in user space (FUSE) to accomplish this. Metadata operations such as file creates and permission changes go through the standard NFS protocol, whereas write operations go through Rapid NFS.

Rapid CIFS is a Windows-certified filter driver that also ensures that only unique data is written to QoreStor. All chunking and hash computations are done at the client level.

Rapid NFS and Rapid CIFS benefits

When Rapid NFS and Rapid CIFS are used with QoreStor they offer the following benefits:

• Reduce network utilization and DMA backup time
  • Chunk data and perform hash computation on the client; transfer chunked hash files on the back-end
  • Reduce the amount of data that must be written across the wire
• Improve performance
• Support DMAs such as CommVault, EMC Networker, and Tivoli Storage Manager. For the current list of supported DMAs, see the QoreStor Interoperability Guide.
• Compatible with existing NFS and CIFS clients — just need to install a plug-in (driver) on the client
  • Can use Rapid NFS and Rapid CIFS to accelerate I/O operations on any client — including a client that uses home-grown backup scripts
  • Can service multiple and concurrent media server backups

Best practices: Rapid NFS

This topic introduces some recommended best practices for using Rapid NFS operations with QoreStor.

• Containers must be of type NFS/CIFS

  RDA containers cannot use Rapid NFS. If you have existing NFS/CIFS containers, you do not need to create new containers to use Rapid NFS; you can install the plug-in (driver) to existing clients.

• The Rapid NFS plug-in (driver) must be installed on client systems
   
  After the plug-in is installed, write operations will go through Rapid NFS while metadata operations such as file creates and permission changes will go through the standard NFS protocol. Rapid NFS can be disabled by uninstalling the plug-in.

• Markers must be set on the client, not in the QoreStor GUI
• If you are using a DMA that supports a marker, should explicitly set it. Your containers should have the marker type of None until you set the marker using the Mount command on the client (after installing the Rapid NFS plug-in).
  • For existing containers, re-set the marker by doing the following:

    For example, if you wanted to set the CommVault marker (cv):

    mount -t rdnfs 10.222.322.190:/containers/backup /mnt/backup -o marker=cv

    Mount command usage:

    rdnfs [nfs mount point] [roach mount point] -o marker=[marker]

    where:

    nfs mount point = Already mounted nfs mountpoint

    roach mount point = A new mount point

    marker = appassure, arcserve, auto, cv, dump, hdm, hpdp, nw, or tsm

• Your QoreStor system must meet the minimum configuration

  Rapid NFS is available on a QoreStor system and a client with a minimum of 4 CPU cores running at a minimum of 4 GHz cumulative processing power and 2 GB memory. Kernels must be 2.6.14 or later. For a list of supported operating systems, see the QoreStor Interoperability Guide. If you update your operating system, you must update your Rapid NFS plug-in as well. Updates are available on the Quest Support site.

• Rapid NFS is stateful

  If the QoreStor system goes down, the connection will terminate. DMAs will restart from the last checkpoint.

• Rapid NFS and passthrough mode

  If Rapid NFS mode fails for any reason, QoreStor falls back to regular NFS mode automatically. For details, see Monitoring Performance.

• Rapid NFS performance considerations

  When using Rapid NFS on your client, Quest recommends that you do not run other protocols to the QoreStor system in parallel, as this will adversely affect your overall performance.

• Rapid NFS acceleration constraints
  • Rapid NFS does not support:
    • Direct I/O memory
    • Mapped files
    • File path size greater than 4096 characters

    • File write locks across clients

  NOTE: If the client and server do not have the same times, the times seen will not match typical NFS behavior due to the nature of file system in user space (FUSE).