[ NOTE: machine translation with the help of DeepL translator without additional proofreading and spell checking ]
[Authors: Marco Hieronymus and Marcel Düssil]
In the course of this blog series, we decided to split the content into two posts. This post focuses on the necessary configurations within Rubrik, VMware and Pure Storage. The second post, on the other hand, will cover backup and restore operations and will provide deep technical insights.
Why Rubrik and Pure Storage?
Pure Storage and Rubrik have revolutionized the industry by modernizing and changing the entire user experience. Infrastructure solutions of the past were complex and inefficient. They lacked the scalability and consistency to deliver what today's applications need. In the past, this only meant increasing costs of your legacy hardware.
With Pure Storage, the solution needed (design goals) to be simple and easy to use, and also include highly efficient software with inline dedupe and compression, replacing the "spinning" mechanical disk entirely with flash (full-flash). But that's not all, it has transformed the customer experience around with Evergreen Storage: Pure Storage guarantees your maintenance costs throughout the system lifecycle. You get free controllers every three years (with Gold Support), all features included, and you never pay twice for storage you've already purchased. Imagine a world where you never have to repurchase the same storage array every 3 to 5 years, never have to migrate again, and never have to deal with downtime for upgrades + updates. That world exists with Evergreen Storage.
Now when you combine Pure with Rubrik, you get a simple and intuitive next-generation data center that eliminates bottlenecks and enables maximum efficiency and cost reduction. Together, the solutions are so simple that you can be up and running within a day. Front-end applications get consistently high performance at an impressive cost model. Plus, your backups are optimized while bottlenecks and complexity are eliminated.
Among other things, both companies follow an API-first strategy. They provide a REST API for your ecosystem to develop and integrate system solutions. For example, on FlashArray we have a highly efficient snapshot solution for application recovery and replication. The API's of both technologies integrate with each other to leverage their snapshot technology and dramatically improve backup performance. The result is simple, policy-driven data management that provides users with multiple consistent copies at a given point in time and with file-level granularity, without application timeouts or impact. This integration also includes FlashBlade with Rubrik, which gives you backup, disaster recovery, replication and data archiving capabilities in the private or public cloud.
At Pure, there are two solutions that make it possible to deploy flash across the enterprise. FlashArray is a flash block product and FlashBlade is Pure's next generation file and object solution. We will still report on this accordingly in due course.
"The combination of Rubrik and Pure Storage enables the highest possible level of automation, with easy administration and constant maximum performance" - Marco Hieronymus (System Engineer for Rubrik)
Personally, I quickly found my way around the Rubrik interface and felt very comfortable. As I have come to expect from Pure Storage, Rubrik has managed to implement simplicity in their product as well. In short, the ecosystem consisting of Pure rubric is well rounded in itself. For me, the only "small challenge"/adaptation to classic backup (my complementary specialties like CommVault and Veeam) is working with SLA domains. With Rubrik, primarily the consideration of when I want to run my backup falls away. The setting when I scheduled my backup is optional here. More on this later.
So we define the main advantages of the Pure Rubric stack as follows:
Deployments in a few hours compared to weeks
superior performance at near zero latency and significantly reduced VM/application in/out impact
unified management of your enterprise data across on-premise and cloud systems
powerful onboard reporting
Reduce footprint by 70% or more
The test environment used in the article consists of:
I recently set up a completely "fresh" test environment based on two nested VMware ESXi hosts in version 6.7. Both hosts have limited resources of 12GB RAM and corresponding 8vCPUs (2x4). To semi-represent a realistic test environment, the nested ESXi host has two VMXNET3 NICs (VMNIC) each. There are two vSwitches, the first vSwitch is used as VM network and the second vSwitch for iSCSi (I know: the sense by the nested operation you can actually do without two vSwitches at this point). Storage resources (Pure) are thus connected via iSCSi. I am aware that the environment is definitely not designed for performance tests, but I must say that this is definitely sufficient for functional tests.
In the test environment, we have 1 Rubrik Brik consisting of 3 nodes with a total of 24 cores and 192GB RAM and 1.2TB SSD/34.5TB HDD of physical storage (1 node = 8 cores, 64GB RAM, 400GB SSD/11.5 HDD). The appliance provides 11.3TB of usable backup storage.
The compute resources are primarily used for high-performance live mounts of VMs and databases. A linear expansion with rubric is given similar to HCI platforms.
It is important to note at this point that the rubric environment is not used exclusively for Pure Storage, but that other systems are also connected and backed up.
As already known, I have two FlashArray X50 with a 19.2TB DataPack each at my disposal. Both arrays are set up accordingly redundant for the storage connection via FC and iSCSi. In this nested ESX environment, I present multiple standalone and mirror volumes to the hosts. The environment is also already in use of VMware VVOLs.
It is important to note that this article is about connecting a Pure Storage array to the rubric. Of course, the classic ActiveCluster operation with Rubrik can also be set up and backed up. Here, only the second system must be set up on the rubric. The logic for the backup is formed by the entire stack.
The configurations at VMware are kept low. Here, only a corresponding backup/service account must be created for the appliance interventions.
In addition, the created user must then be linked accordingly with the respective level/entity and the role. We assigned unrestricted administrator rights to all objects here.
For productive environments this procedure is rather suboptimal and it is recommended to create "slimmed down" and adapted roles with the corresponding mandatory necessary rights. These rights can be found officially in the Support Portal and in article 000003318 "Minimum VMware Permissions for Rubrik CDM integration".
Preparation Pure Storage
Also on the Pure Arrays only one service account has to be created for setting up the API integration with "storage admin" rights. An API key does not need to be registered manually on the FlashArray.
We document the created service account for later use or reuse in case of adjustments.
In order to successfully connect to the REST API, it is important to note that each of the rubric nodes here must have access to port 443 in case of network restrictions. A corresponding list of the required ports is given below:
The systems to be connected and already prepared must now also be configured accordingly for API use on the rubric. To do this, we log in to the Web GUI section. After successful login, we land on the dashboard.
First, we integrate the Pure Storage Array via the "gear" in the header.
We notice that in Rubrik OS 5.1, Pure Storage is exclusively available for selection as "Array Type". As of today, Rubrik only offers native storage integration for Pure Storage. We can neither confirm nor deny whether this is due to the above common features of both solutions (design/philosophy principle). In my eyes, however, it is obvious that the development of Rubrik has therefore preferred Pure Storage.
We only have to specify the "virtual IP" and the previously set up credentials of the Pure Storage array to be connected. By clicking on "Add", the connection is initiated, checked and the API key is exchanged in the background. This process and operations can also be seen in the Pure audit or session trail. Rubrik checks in regular intervals (5 minutes) if the connection is working properly.
Next, we have to set up the vCenter connection similar to the storage arrays. This is also done via the settings in the header. We switch to "vCenter Servers".
The IP address (ideally FQDN in productive environments) and the access data of the prepared service account are stored here. The advanced settings can be neglected at this point. An interesting feature is the "Automatically link discovered virtual machines" checkbox activated by default, which then directly triggers an inventory of the vCenter objects.
The setup or the status of the subsequent inventory can be seen in the Activity Log.
Rubrik works with so-called "local domains".
As mentioned, rubric systems need to be rethought away from classic backup jobs to SLA jobs. Virtual machines are divided into SLAs. This allows flexibility in backing up machines on a snapshot basis continuously and automatically throughout the day.
With Rubrik, it is also possible to link one system to another Robo system at the remote site, this allows you to manage Local SLA domains and Remote SLA domains within one system. Since we are only on the local system in the current setup, we set up a new Local Domain.
We assign an SLA name here and can select the criteria and intervals for the backup. So we backup once a day and keep this backup for 7 days. Optionally, a corresponding time window can also be set up.
Now we still need to link the backup objects to the Local Domain.
To use the previously set up storage integration with Pure for backups, this must also be enabled for the respective VMs. This is "VM-based" and can be defined individually for each machine. This setting can be made within the backup object/VM via the "..." button with "Enable Array Integration". Just a one-click.
It is important to enable Storage Array Integration at the point before assigning the SLA/Local Domain. Otherwise, a snapshot is immediately created on a streaming backup basis (classic VM backup - if no backup time window is stored) when the virtual machine is assigned.
The local domain must then be linked to the VM. This is done via "Manage Protection".
All active and past activities can be displayed on the whole in the Activity Log or per object e.g. within a VM.
HINT: It is possible to search for objects using the search bar in the header. In my eyes, ideally solved for quick searching.
Optionally, Rubrik does not withhold the possibilities of features like VM tagging. It is also possible to automate assignment to individual local domains. A VM administrator could automatically assign backup SLAs based on tags set up and do so without additionally accessing the Rubrik interface. Personally, I'm a big fan of such features as they bring additional simplicity and automation.
Setup on Rubrik page is done through the Virtual Machine tab > Tags > SELECT vCenter. Rubrik also inventories existing tags, allowing for easy mapping with existing Local Domains.
The software is also smart enough to recognize when an assignment has already been made to an object (here a VM) (see screenshot).
That's it for the first part of our blog series on "An easy and fast way to backup FlashArray with Rubrik". We think that the integration of Rubrik is coherent and round. The ecosystems complement each other and you can work very chic and especially easy in combination with both systems.
Marco and I have known each other for some time. We are connected not only by friendship but also by our profession. Marco has years of expert knowledge in VMware, Nutanix, DELL EMC VxRail/Networking, Aruba/Cisco Switching and Rubrik, among others. He has several certifications to his credit in his areas of expertise. In the course of his career, Marco was not only employed in the system house, but also for a few years as an end customer at the regional daily newspaper in Regensburg. In the course of my training I was able to pass my technical paper with flying colors thanks to Marco's help, among other things.
At this point I would like to thank you again for your support and time - Marco.
More info - Links
All officially published setting options in the GUI but also CLI can be read via the "on-board" user guides of the Pure Storage systems.
Click on "Help" in the Purity main menu.
The User Guide is structured like the main menu and can be opened downwards. A search function is also integrated - within here you can also search for keywords.
WEB: Pure Storage (Pure1) support portal - Ticket system and support *(requires registered FlashSystems)
PHONE: Pure Storage phone support: GER - (+49) (0)800 7239467; INTERNATIONAL - (+1) 650 7294088