Data Domain: Expedited configuration steps and details for Data Domain Boost File System

The Data Domain BOOST FS is a new feature available with for DDOS 6.x and later by which customers may leverage the BOOST libraries to access a remote Data Domain storage unit by locally mounting on any supported Linux server, without the need for the backup application to have a custom BOOST library or plug-in.

This feature lets supported customers' applications to drop backup files to a local Linux directory (where the remote DD storage unit is mounted) and the files to be stored in the DD, with most of the advantages of BOOST, such as deduplication performed on the client (hence reducing the needs for bandwidth), and with no need for the application to be changed at all.

BOOST FS is released for the first time to customers with DDOS 6.0. The official documentation to learn more and initially configure it is Data Domain BoostFS 1.0 Configuration Guide.

Despite the official documentation containing all the updated details, BOOST FS on its initial release only supports the following. It may be used with other versions, but DELL cannot guarantee that it shall work as expected:

• DDOS 6.0
• Older Red Hat Enterprise Linux, SUSE Linux Enterprise Server, and Ubuntu release
• Linux FUSE 2.8 (FUSE stands for Filesystem in User Space)
• A few applications on the Linux side:
  • CommVault Simpana versions 9, 10, and 11
  • MySQL Community 5.6. and 5.7
  • MySQL Enterprise Manager 5.6 and 5.7
  • MongoDB Community 2.6, 3.0, and 3.2

Configuration of FUSE requires both Data Domain side and Linux client-side configuration:

• On the Data Domain side, there must be a particular Storage Unit created with some BOOST user assigned for the purposes of storing data coming from BOOST FS
• On the Linux client side, the FUSE libraries (that come with the OS) and the BOOST FS (provided by DELL) must be installed, and some configuration performed so the Linux operating system can see and mount the remote Data Domain storage unit locally in some Linux client mount point

On the Linux client, there is a requirement to set up the way to authenticate to the Data Domain, which can be either "self-contained" (through the lockbox), and by using either an existing Windows Active Directory or UNIX or Linux based Kerberos Key Distribution Center. Details for the kerberos-based authentication configuration are outside of this document, and customers should see the official documentation.

The steps below are a condensed, executive summary of the configuration steps in the guide, see the official documentation for all the details.

Data Domain side configuration for BOOST FS

BOOST distributed segment processing (DSP) must be enabled for BOOST FS to work. Check if this is so through:

# ddboost option show Option Value ------------------------------ ------- distributed-segment-processing enabled virtual-synthetics enabled fc enabled ------------------------------ -------

A particular user must be created for using the BOOST FS storage unit to be created, and the user must be assigned to that storage unit:

#### Create BOOST FS user # user add boostfsuser role admin password abc123 User "boostfsuser" added. 
#### State this new user is for use with BOOST # ddboost user assign boostfsuser User "boostfsuser" assigned to DD Boost. 
#### Create a new storage unit named "BOOSTFS_LSU" and assign the new user to it # ddboost storage-unit create BOOSTFS_LSU user boostfsuser Created storage-unit "BOOSTFS_LSU" for "boostfsuser".

In the end, this is how the storage unit should show in the Data Domain CLI:

# ddboost storage-unit show Name Pre-Comp (GiB) Status User Report Physical Tenant-Unit Size (MiB) -------------------------------------- -------------- ------ ----------- --------------- ----------- ... BOOSTFS_LSU 0.0 RW boostfsuser - - -------------------------------------- -------------- ------ ----------- --------------- -----------

Linux client-side configuration for BOOST FS

After checking the Linux operating system is one of those supported, get the up to date BOOST FS operating system package. Here we used an Ubuntu 14.04 as an example. Hence, downloaded the operating system .deb file, and install it from the command line after checking all dependencies are satisfied:

# ls -l -rwxrwxr-x 1 root root 2429502 May 26 05:01 DDBoostFS_1.0.0.4_546961_amd64.deb # dpkg --dry-run -i DDBoostFS_1.0.0.4_546961_amd64.deb Selecting previously unselected package ddboostfs. (Reading database ... 465833 files and directories currently installed.) Preparing to unpack DDBoostFS_1.0.0.4_546961_amd64.deb ... # dpkg -i DDBoostFS_1.0.0.4_546961_amd64.deb Selecting previously unselected package ddboostfs. (Reading database ... 465833 files and directories currently installed.) Preparing to unpack DDBoostFS_1.0.0.4_546961_amd64.deb ... Unpacking ddboostfs (1.0.0.4) ... Setting up ddboostfs (1.0.0.4) ... Processing triggers for libc-bin (2.19-0ubuntu6.7) ...

The package files are all installed below the /opt/emc/ path, including the bin/ subdirectory, and the location of the plug-in logs (log/):

# ls -rtl /opt/emc/ drwxr-xr-x 9 root root 4096 May 26 05:01 boostfs # ls -rtl /opt/emc/boostfs/ drwxrwxrwx 2 root root 4096 Oct 7 2016 log drwxrwxrwx 2 root root 4096 Oct 7 2016 lockbox drwxr-xr-x 4 root root 4096 May 26 05:01 man drwxr-xr-x 2 root root 4096 May 26 05:01 lib drwxr-xr-x 2 root root 4096 May 26 05:01 etc drwxr-xr-x 2 root root 4096 May 26 05:01 bin

Notice that the owner and group for the installed files may differ from those shown here. The next step is to configure the authentication, although there are several methods for doing so, only the self-contained "lockbox" method is explained here. To do so, and for all other BOOST FS configuration details on the Linux side, the "/opt/emc/boostfs/bin/boostfs" command is used. BOOST FS also has a text configuration file, but the details should be checked in the official documentation.

The "lockbox" must be used to authenticate this Linux client against the remote Data Domain:

# /opt/emc/boostfs/bin/boostfs lockbox -h Usage: boostfs lockbox set -u <storage-unit-username> -d <data-domain-system> -s <storage-unit> boostfs lockbox {remove | query} -d <data-domain-system> -s <storage-unit>

For this example, we use a command such as the following, entering the DD BOOST FS user password when prompted:

# /opt/emc/boostfs/bin/boostfs lockbox set -u boostfsuser -d REMOTE_DD_IP_ADDRESS -s BOOSTFS_LSU Enter storage unit user password: Enter storage unit user password again to confirm: Lockbox entry set

At this stage, we should already see the remote Linux client as registered for BOOST from the DD CLI:

# ddboost show connections Active Clients: 0 Clients: Client Idle Plugin Version OS Version Application Version Encrypted DSP Transport ------------------------------ ---- -------------- ------------------------------------- -------------------------------------------------- --------- --- --------- LINUX_CLIENT_NAME YES 3.3.0.2-545054 Linux 3.13.0-71-generic x86_64 BOOSTFS:1.0.0.4-546961 Test Application FUSE:2.9.2 YES YES IPv4 ------------------------------ ---- -------------- ------------------------------------- -------------------------------------------------- --------- --- ---------

At any time, the lockbox status can be checked with:

# /opt/emc/boostfs/bin/boostfs lockbox query -d REMOTE_DD_IP_ADDRESS -s BOOSTFS_LSU Lockbox entry is present for storage unit user=boostfsuser

Finally, the remote Data Domain storage unit may be mounted locally as any other Linux file system locally. The "app-info" is used here as an example, it should be a string that identifies the particular application to use this mount point for logging:

#### Create mount point on Linux server # mkdir /opt/boostfs_mountpoint/ #### Mount the remote DD storage unit # /opt/emc/boostfs/bin/boostfs mount -d 10.60.29.233 -s BOOSTFS_LSU -o app-info="Test Application" /opt/boostfs_mountpoint/ mount: Mounting 10.60.29.233:BOOSTFS_LSU on /opt/boostfs_mountpoint/ #### We can see the remote storage unit is mounted, and the details match those obtained from the DD "df" command line # df -h /opt/boostfs_mountpoint Filesystem Size Used Avail Use% Mounted on boostfs 64T 512G 64T 1% /opt/boostfs_mountpoint # filesys show space Active Tier: Resource Size GiB Used GiB Avail GiB Use% Cleanable GiB* ---------------- -------- -------- --------- ---- -------------- /data: pre-comp - 33096.9 - - - /data: post-comp 65460.3 511.7 64948.6 1% 0.0 /ddvar 29.5 19.9 8.1 71% - /ddvar/core 31.5 0.2 29.7 1% - ---------------- -------- -------- --------- ---- --------------

Any application can copy or write any files under the /opt/boostfs_mountpoint/ path, and those files are transparently and sent over the wire to the remote Data Domain storage unit, leveraging deduplication performed on the source end (on the Linux server) rather than on the DD host, and hence reducing the amount of data sent over the wire.

When access to the remote Data Domain storage unit is no longer necessary, the remote storage unit may be unmounted by running:

# /opt/emc/boostfs/bin/boostfs umount /opt/boostfs_mountpoint/