FLASHBLADE™ AS AN

ARCHIVAL TIER FOR RUBRIK

Backup, Restore, and Long Term Data Archiving
November 2017
TABLE OF CONTENTS
INTRODUCTION .................................................................................................................................... 2
TEST ENVIRONMENT ........................................................................................................................ 2
TEST OVERVIEW ................................................................................................................................... 4
RUBRIK, FLASHBLADE, AND FLASHARRAY INTEGRATION SETUP ...................... 4
RUBRIK SLA POLICY CREATION AND EXECUTION ........................................................ 8
VM RECOVERY FROM RUBRIK TO FLASHARRAY .......................................................... 14
CONCLUSION ........................................................................................................................................ 18

©2017 Pure Storage, Inc. 1

INTRODUCTION
The purpose of this whitepaper is to showcase the ease of integration, best practices, and
expected performance results for data backup, restore, and long term data archiving using
the Pure Storage FlashArray, Rubrik, and Pure Storage FlashBlade™ solutions with realistic
real-world examples. The combination of these innovative products provides industry-
leading performance, high levels of resiliency, simple implementation, and, perhaps most
importantly, low cost per GB.
Three core pieces of next-generation storage technology will be highlighted in this
document and explanations will be provided for how they seamlessly work together:
• FLASHARRAY Delivers high performance primary block storage for the most
demanding datacenter requirements. FlashArray provides built-in snapshots and
replication to other Pure Storage arrays.
• RUBRIK Next-generation appliance for hosting backup data, providing instant
VM/file recovery, as well as orchestrating archival data movement from FlashArray
to FlashBlade.
• FLASHBLADE A fast and dense unstructured data platform from Pure Storage that
provides rapid backup and recovery of archival data, along with other primary use
cases for unstructured data.

TEST ENVIRONMENT
The test cases shown here are intended to emphasize the high levels of throughput,
workload consolidation, minimized RTO and RPO, and ease of administration that these
connected solutions can provide. Though we will focus on virtual machines in our
examples, the use cases shown here are easily extensible to other workloads, such as
SQL databases, VSI, VDI, and Oracle, just to name a few.
Our test environment was comprised of the following elements:
• Five ESXi 6.5 hosts running 250 Windows 10 desktops with 50GB drives (27GB
used). The desktops included MS Office 2016, Adobe Reader, and numerous iso, pdf,
mp4, and many other pre-compressed, commonly used files. Each ESXi host features
two redundant 10GB network connections and two redundant 16GB Fibre Channel
HBA connections
• One Pure Storage FlashArray//M20 with 10TB RAW storage for primary storage (up to
25 TB effective capacity assuming a 5:1 data reduction ratio, which are typical results
shown from the Pure Storage install base)
• Rubrik r348 Four Node Cluster for secondary storage backup and recovery
orchestration between FlashArray and FlashBlade
• One Pure Storage FlashBlade with seven blades (a half-populated chassis) for the
archival tier, with 8TB per blade
• Two Brocade VDX6740T Switches for 1/10/40GB Networking
• Two Cisco MDS 9148S 16GB Fibre-Channel Switches

©2017 Pure Storage, Inc. 2

The simplified connectivity diagram below illustrates how the various solutions are
integrated with one another and how there is no single point of failure within the design:

10Gb iSCSI Connectivity

Spine
WAN
16Gb FC Connectivity Network

10/40 GbE NW Connectivity

Brocade VDX6740T - A
IO IO I IO IO I
Brocade VDX6740T - B
Brocade Brocade
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
VDX 6740T Brocade Trunk 10GE 40GE/FC
VDX 6740T Brocade Trunk 10GE 40GE/FC

Archival Archival
Data Data

Backup FlashBlade Backup

Data Data
Archival Storage

Rubrik r348 Four Node Brik

Secondary Storage

Network Network
2 1

RE SE T

R O M
SUPERMICR �
SUPERMICR
�

2 1

RE SE T

R O M
SUPERMICR �
SUPERMICR
�

2 1

RE SE T

R O M
SUPERMICR �
SUPERMICR
�

Prod. Prod.
ESXi 6.5 Servers
Data Data

DS-C9148S-K9 DS-C9148S-K9
CONSOLE

CONSOLE

USB USB
STATUS STATUS
MGMT ETH

MGMT ETH

P/S P/S

FAN FAN
LINK ACT LINK ACT
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48
MDS 9148S 16G Multilayer Fabric Switch MDS 9148S 16G Multilayer Fabric Switch

Cisco MDS 9148S - A Cisco MDS 9148S - B

Prod. Prod.
Data Data

FlashArray//M20 R2
Primary Storage

Figure 1. High-level architectural diagram of solution

©2017 Pure Storage, Inc. 3

TEST OVERVIEW
The 250 Windows 10 desktops were hosted on 5 ESXi SuperMicro servers running under
a common vSphere 6.5 instance. Pure Storage FlashArray features a fully-integrated
vCenter plug-in that enables single pane of glass administration of the primary storage
array. For our testing, Rubrik leverages VMware VADP technology to integrate with
vCenter and provide CBT based backup capabilities for all virtual machines running
within the vCenter instance.
Our test process will cover the following steps, with instructions and explanations for
each phase:

Figure 2. Test process to be executed

RUBRIK, FLASHBLADE, AND

FLASHARRAY INTEGRATION SETUP
Connecting vCenter to the Rubrik appliance is easily accomplished by registering it as a
source, as shown in the next few screenshots.

©2017 Pure Storage, Inc. 4

Under the Protection menu select Sources.

Figure 3. Selecting sources in Rubrik GUI

Figure 4. vCenter registration for Rubrik

Fill out the credential information for your vCenter instance. The account used to register
the vCenter source to Rubrik must have administrative privileges.

©2017 Pure Storage, Inc. 5

Figure 5. vCenter registration wizard

We next setup Rubrik to serve as the connection and orchestration point between the
Pure Storage FlashArray (primary storage) and the Pure Storage FlashBlade (archival
storage) layers via the Rubrik archival tier option. FlashBlade enables extremely fast and
secure backup and recovery of information, which is hugely important in the healthcare
and high-performance computing industries but also extends into other businesses.
FlashBlade offers inline compression and massive density (1.5PB in 4U effective) – enabling
your datacenter footprint to shrink while also minimizing power and cooling costs.
Setting up the FlashBlade archival tier was straightforward. First, we name, size and set
our export rules and permissions for our File System on the FlashBlade GUI:

Figure 6. File System creation on FlashBlade

©2017 Pure Storage, Inc. 6

With our FlashBlade File System created, the next step was to connect it to the Rubrik
appliance as an archival tier.
Select the Archival Locations option from the Rubrik menu located under the gear on the
top right.

Figure 7. Registering FlashBlade File System as an archival location in Rubrik

Click on the + sign to add the FlashBlade archival tier, entering a data IP address and
directory name, as shown in the below example.

Figure 8. Wizard for registering FlashBlade as a Rubrik archival location

©2017 Pure Storage, Inc. 7

Completing the wizard will connect Rubrik to the FlashBlade archival tier. An overall
summary can be seen below:

Figure 9. Summary screen of FlashBlade archival location in Rubrik

RUBRIK SLA POLICY CREATION

AND EXECUTION
Once vCenter was connected to the Rubrik appliance, we generated a Protection Policy
that would allow for daily recovery of any single or group of virtual machine(s) from our
test group or even any single file or group of files within a selected desktop. The first step
is to create a protection policy that dictates the frequency and retention of VM snapshots,
as well as other options.

©2017 Pure Storage, Inc. 8

Expand the SLA Domains menu and highlight the Local Domains option.

Figure 10. Rubrik Local Domains SLA creation

In the top-right of the GUI, click on the + sign that is spawned to get into the SLA domain
creation wizard.
For our test dataset, we will be taking a snapshot of our 250 desktops every 12 hours and
keeping them for 3 days. Monthly backups will also be created and retained for one year.
Windows for when you want to run or not run the protection job can also be assigned
here, as well as setting replication between multiple Rubrik appliances and our archival
tier, which we will cover next.

Figure 11. Rubrik SLA creation wizard

©2017 Pure Storage, Inc. 9

With the local protection policy defined, we next click on Configure Remote Settings in
order to add the FlashBlade archival tier.

Figure 12. Configure Remote Settings in Rubrik SLA

Enable the Archival slider button and select the NFS archival tier for FlashBlade that
we created a few steps ago. For our testing, we also checked Enable Instant Archive to
maximize bandwidth between Rubrik and FlashBlade.

Figure 13. Setting up FlashBlade archival tier within Rubrik SLA

Next, our 250 VMs were selected and the SLA created in the previous step was
assigned to them.
Finally, we selected the 250 VMs from our vCenter instance. The desktops were
segregated into five folders containing approximately 50 VMs each. The following
screenshots depict how they were selected.

©2017 Pure Storage, Inc. 10

Figure 14. Selecting 250 test VMs within Rubrik GUI

After highlighting the 5 VM folders that each contain 50 desktops, we selected

Manage Protection.

Figure 15. Selecting the 250 desktops to be protected from our vCenter source

To complete setup for protecting the 250 VMs, select the SLA we created earlier and
click Submit.

©2017 Pure Storage, Inc. 11

Figure 16. Assigning the SLA for 250 test VMs

During this initial protection run for our 250 VMs, we observed high transfer speeds
between FlashArray and the Rubrik cluster and, later, to the FlashBlade archival tier.
A screen capture of the VMs being ingested from the Pure Storage FlashArray GUI to
the Rubrik appliance can be seen below.

Figure 17. 250 Virtual Machines being ingested from FlashArray to Rubrik

Worth noting from this experiment is that the initial VM ingest will always be the most
“expensive” in terms of bandwidth, IOPS, and time. That is, subsequent backup runs will
only update the changes made to the target VMs, rather than running a full VM backup,
as was done in this first example.

©2017 Pure Storage, Inc. 12

Moving from Rubrik to the FlashBlade archival tier introduces a new variable that will
allow granular control over your workload archiving. Most important is that the number
of VMs that can be sent to FlashBlade from Rubrik is a tunable value that Rubrik support
can change based upon the customer’s requirements, available network bandwidth, and
other factors. By default, each Brik will offload one VM at a time to the archival tier, but
this number can be increased dramatically by working with your Rubrik support team.
As an example of the increased archival settings in action, the FlashBlade GUI screenshot
below shows a portion of our 250 VM job being written with the Rubrik archival settings
increased to 2 VMs/Brik (8VMs in parallel) from 1VM/Brik.

Figure 18. Virtual Machine ingestion at 2VMs/Brik

Once this job completed we removed all archived VMs from FlashBlade and worked with
Rubrik support to further increase this value to 10VMs/Brik or 40 VMs in parallel. In the
next screenshot we can see more consistent sustained write performance, higher peak
write bandwidth, and that the initial archive job completed in just over 6 hours. This is
several times faster than using the default Rubrik archival settings.

©2017 Pure Storage, Inc. 13

Figure 19. Virtual Machine ingestion at 10VMs/Brik

This proves that Rubrik and FlashBlade have the flexibility and performance needed to
backup and restore your most critical infrastructure components according to whatever
requirements are dictated by your business.

VM RECOVERY FROM RUBRIK

TO FLASHARRAY
Rubrik offers multiple methods for recovering a VM or multiple VMs in parallel. For our
example, we will show the steps needed to recover an entire VM or individual files or
directories from within the Rubrik GUI. However, multiple VMs can easily be restored
in parallel using Rubrik’s REST API. Please see the following blog post for scripting
examples of how to restore multiple VMs in parallel:
https://virtuallysober.com/2017/05/15/scripting-a-rubrik-recovery-plan-using-rest-apis-
powershell/
Back to our example. First, we will select our target VM to be recovered. VMs can be
searched from within the Virtual Machines section of the GUI or from the primary search
bar across the top of the Rubrik UI. For this test, we chose the Win10-10 desktop and
clicked on the name.

©2017 Pure Storage, Inc. 14

Figure 20. Selecting a test VM to be restored from Rubrik GUI

Clicking on the VM name takes us to a summary screen of the VM’s protection. The blue
dots on the calendar indicate available snapshots; clicking on one will show snapshot
timestamp(s).

Figure 21. Detail pane for VM to be restored showing snapshot metrics in the Rubrik GUI

The snapshot in question has been archived to FlashBlade, but a copy also exists locally
on Rubrik. When the icon highlighted below is not available, only the FlashBlade archival
backup is available for a given snapshot. Snapshots that only reside on FlashBlade will
need to be imported to the Rubrik appliance before the VM or file can be recovered, but
that operation is accomplished very quickly.
To select a recovery option for the snapshot, click on the highlighted icon below.
(Also worth noting is that individual files or directories within the virtual desktop can
be restored either via search or drilling down into the hard drive of the VM.)

©2017 Pure Storage, Inc. 15

Figure 22. Selecting the Rubrik snapshot from which to restore the VM

That will open the menu to show the available restoration options for the VM.

Figure 23. Snapshot restore options available within the Rubrik GUI

In our test, we are going to use the Export option. This option will instantly recover the
snapshot onto the Rubrik appliance, boot the desktop so that it can immediately start
serving IO, and then transparently vMotion it back to a location of our choosing on the
primary Pure Storage FlashArray. Setting the destination host and FlashArray datastore
occurs in the Export wizard that spawns in the next screenshot. In this example, we will
completely overwrite the existing VM with this earlier snapshot using the same ESXi host
and FlashArray datastore.

©2017 Pure Storage, Inc. 16

Figure 24. Selection of ESXi host, datastore for snapshot export wizard

Once initiated, we can see that the virtual desktop restoration begins immediately.
Progress is tracked within the Rubrik GUI.

Figure 25. VM export restoration from snapshot

The virtual machine is booted from the snapshot on the Rubrik appliance and
transparently vMotioned over to the Pure Storage FlashArray from the Rubrik appliance.
This operation completed in approximately 12 minutes, with the VM serving IO within 3
minutes through Rubrik.

©2017 Pure Storage, Inc. 17

CONCLUSION
Through this simple demonstration, we have proven that the combination of FlashArray,
Rubrik, and FlashBlade delivers the entire suite of data performance, protection, and
agility. Additional workloads can easily be mixed and managed from the interfaces
shown in this guide across the entire business. As data capacity requirements grow in
both primary and secondary use cases, both FlashBlade and Rubrik can non-disruptively
scale and be upgraded – a process that is completely transparent to customers. Finally,
systems administrators can move away from tedious and repetitive tasks that are focused
on keeping their infrastructure online and instead move up the stack to concentrate on
tasks that will improve the company as whole.

© 2017 Pure Storage, Inc. All rights reserved. Pure Storage, the “P” Logo, and FlashBlade are trademarks or
registered trademarks of Pure Storage, Inc. in the U.S. and other countries. Rubrik is a registered trademark
of Rubrik in the U.S. and other countries. The Pure Storage product described in this documentation is
distributed under a license agreement and may be used only in accordance with the terms of the agreement.
The license agreement restricts its use, copying, distribution, decompilation, and reverse engineering. No part
of this documentation may be reproduced in any form by any means without prior written authorization from
Pure Storage, Inc. and its licensors, if any.

THE DOCUMENTATION IS PROVIDED "AS IS" AND ALL EXPRESS OR IMPLIED CONDITIONS,
REPRESENTATIONS AND WARRANTIES, INCLUDING ANY IMPLIED WARRANTY OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE, OR NON-INFRINGEMENT ARE DISCLAIMED, EXCEPT TO THE
EXTENT THAT SUCH DISCLAIMERS ARE HELD TO BE LEGALLY INVALID. PURE STORAGE SHALL NOT
BE LIABLE FOR INCIDENTAL OR CONSEQUENTIAL DAMAGES IN CONNECTION WITH THE FURNISHING,
PERFORMANCE, OR USE OF THIS DOCUMENTATION. THE INFORMATION CONTAINED IN THIS
DOCUMENTATION IS SUBJECT TO CHANGE WITHOUT NOTICE.

ps_wp_flashblade-as-archive-for-rubrik_02

©2017 Pure Storage, Inc. 18

Pure Storage, Inc.
Twitter: @purestorage
www.purestorage.com

650 Castro Street, Suite #260

Mountain View, CA 94041

T: 650-290-6088
F: 650-625-9667

©2017 Pure Storage, Inc.

Sales: sales@purestorage.com 19