A New Standard for Backups – Rubrik
June 29, 2015
It’s pretty weird to get excited about backups, but I’ve found myself thinking how cool the new technology that Rubrik‘s designing. If you haven’t heard of these guys yet, you will. They presented at Virtualization Field Day 5 in Boston and had some new announcements that will blow your socks right off your feet.
All travel expenses and incidentals were paid for by Gestalt IT to attend Virtualization Field Day 5. This was the only compensation given and did not influence the content of this article.
The first of these announcements really had nothing to do with technology, but rather with people. Rubrik announced to the world that my good friend Chris Wahl was joining their team as Technical Evangelist. Thats a pretty big “get” for Rubrik, signing a top 10 Virtualization blogger and two times over VCDX to their team. I think it also lends some credence to their legitimacy, as I don’t suspect Mr. Wahl would have joined a team that didn’t already have something going for it.
Why is Rubrik Cool?
The Rubrik CEO Bipul Sinha explained to us that the main objective of Rubrik was to make backups useful and easy to manage. His example was using Apple’s Time Machine technology. How come my laptop backup is so easy to manage, but a corporate server backup is so difficult. Obviously servers are backed up more often with more stringent SLA etc, but his point was well received. It should be easy to do because no one wants to spend their time managing backups.
Rubrik’s model uses a hardware appliance with a scale out architecture similar to Nutanix. (This shouldn’t be a coincidence since Bipul is a founding investor in Nutanix) Rubrik’s appliance is sold as a brick (chassis) with 4 nodes (servers) in it. Once the appliance is racked, cabled and IP’d the next step is to connect it to your vCenter server(s). The Rubrik software scans the vCenter for a list of virtual machines and from there, you can select VMs to backup to the Rubrik appliance. Rubrik’s team is touting this process that only takes 15 minutes to get up and running. If you’d like to see more info on this, check out this post from Brian Suhr.
Note: right now this is a 1.0 product and only supports VMware, but the roadmap is to support Hyper-V and Physical machines as well. It is not meant to be only for VMware environments for ever.
Backup Process
The process of backing up virtual machines to the Rubrik device consists of selecting a virtual machine and selecting a SLA. There are some default SLAs that come with the appliance but the backup admin is allowed to create as many as he/she needs in order to meet the organizations retention periods and backup windows.
Once the backups start, the virtual machine is snapshotted and the bits are shipped over to the Rubrik appliance where they are inline deduplicated and stored on flash temporarily. Depending on the SLA, these backups will be stored on disk as well as possibly shipped off to an S3 storage endpoint, most likely Amazon S3. This is neat right? How many times have you heard corporations state that they want to keep all data for seven years right up until they hear how much storage they are going to need to buy to accomplish that? Now Rubrik can keep the most recently backed up information locally on disk but ship off some of the bits to a cloud storage device.
Recovery Process
OK, the backup process is pretty slick. Pick a VM and a policy and let Rubrik do its thing. But everyone knows that the backup is only as good as the recovery process. Rubrik’s recovery model is great! From the Rubrik HTML5 web portal, pick the VM and a backup date to restore, or for a single file restore pick the vm, the file and the file date to restore. Simple process and the search process is VERY fast. This is because all of the backup metadata is stored on the Rubrik flash drives for quick recalls.
Now that we’ve found the files, we can perform either a recovery or an instant mount. The recovery process will power down the existing virtual machine and recover the backup in its place. Nothing new there, but if we need a recovery to take place faster, we can mount the backup directly on the Rubrik flash tier and mount it to vCenter over an NFS mount point.
Summary
I don’t know what these appliances are going to cost, but Bipul assured us all that they won’t disappoint. With an easy backup process, simple and fast recovery process, ability to scale out and still keep a single deduplication domain and a fast storage appliance, all I can think of to say is to “shut up and take my money.” We’ll see how this product does on the market , but I have a feeling that this is going to be the new gold standard for backup solutions.
it would be interesting to see the difference between rubik and veeam 8.
I don’t see anything new and “new standard” about this vs Veeam’s. As for de-duplication, I am using commodity hardware and a number of cheap SATA drive and Windows 2012 R2 Dedup; and I am seeing up to 80% dedup ratio. If I want to scale out, I’ll add some more hard drive, chassis, etc. And oh, my commodity hardware are at a fraction of what the vendor will be selling.
Maybe the product is a “new standard” because it’s so easy to use that grandma can do it? Well, IT folks are smart enough to use Veeam just fine.
Anyway, if you have a chance, please compare it with Veeam or other products out there; and let’s see what makes this a “new standard”
Howard,
Obviously time will tell with this appliance. Veeam has been a great solution for many years now and can’t be considered the new backup solution on the block anymore. There will always be challengers.
I think that making things easier to use is always the goal right? Veeam came into the industry and made difficult to manage backup tasks pretty simple and it sold very well. I have to imagine that an appliance that makes backups very simple will do the same thing.
I’m sure there will be plenty of posts coming that compare Veeam to Rubrik but remember that one is an appliance and the other is only software. There will obviously be differences between them.
Instead to do a compare to Veeam I’d rather get more details on things like the possibility to use this “storage-backup” appliance as principal infrastructure storage, or an additional one, for some workflows…. Things would probably be more clear if we had some numbers, like what’s the number of IOPS we can get out of it. Or this isn’t the right way to use the system?
If it’s just to run backups (even the cool way), the overall price point might set back quite a few folks IMHO.
Second good point is certainly the cloud archiving, where you can pull back just what you need – not a full VMDK.. to do a restore.
@Eric – so basically Rubrik copied some patented Veeam technologies like Instant VM Recovery (ironically first presented also at Tech Field Day – but 5 years ago), and told you this was some new breakthrough? I cannot believe there was no one sitting in the same room with you to bring this funny fact up.
Regarding the appliance making things even simpler, it is true – but in that perspective, how Rubrik is different from many Veeam-based appliances that Veeam partners have been selling for years now? Those are based on industry-standard servers, so are cheap as dirt – and yet can scale to hundreds of TB per single node with servers like Cisco C3160. Also, they are custom tailored to specific customer’s needs and use cases in terms of IOPS, which is very important – just as Vladan notes.
Honestly, but I hardly see anything new here… just another Veeam copy cat on the block with potential legal issues on the horizon for violating Veeam patents. Seriously, what’s the big deal here, in your opinion?
@Vladan – actually Veeam does granular item-level restores directly from the cloud for a while now with its Cloud Connect, so nothing new here either.
Look don’t get me wrong here, I’ve been a Veeam fan for a long time and love the product. I know that Rubrik and Veeam both do this instant recovery thing and make selecting VMs for backup pretty easy to do. I even think that Veeam has a little more flexibility because you can select different repositories and do replication etc. The patents battle, is not my problem and customers will only care if they might lose their investment in a backup technology because of a lawsuit. All that matters is the solution and in this case we’re talking about a primary storage device with easy to manage backup capabilities and it can run the backup workloads directly without having to move them to another tier 1 storage device. It also makes it very simple to move backups to the cloud for even longer retention periods.
To be honest, so far I’ve seen so many Veeam folks upset about Rubrik’s solution that I know its got something going for it. If this was just another backup solution no one would care. Time will tell on this one for sure, but I think there is something here to pay close attention to.
P.S. Just for fun 🙂
http://www.virtualpro.co.uk/2010/07/15/gestalt-it-seattle-tech-field-day-%E2%80%93-day-1-summary/
Solving Rubrik’s “Brik”… what’s new here?
Rubrik communicates with vSphere using Vmware API’s so any IP infringement seems to be based upon overlapping marketing features
with any other another VM backup product. So, the VM data management standard features are not grounds for IP infringement.
Rubrik is agentless for there’s no code added to the vSphere environment. The GUI for backup control runs on the Rubrik cluster.
(Sorry if my reading of the product details is inaccurate).
The Rubrik uniqueness includes:
a Scaleout compressed/deduped file system (combining Isilon and Data Domain feature sets) – one copy of any block across all backed
up VM’s – so only unique blocks drift up into the AWS S3 store over time.
Metadata management that indexes “files” quickly by file name in any of the backed up VM’s and time stamped snapshots.
Google-like search features to find that one unique file you stored 6 months ago on some random VM’s vmdk and later deleted.
Significant CPU, RAM and Flash resources are provided in a “brik” to manage, search and snapshot metadata for quick results.
This might provide a smart tool to examine file “usage” patterns inside VM’s and enable you to recover primary storage by
keeping VM’s lean and knowing you have archived versions of any deleted production files.
NFS exposed VMware mountable target for Dev/QA use cases or zero-copy data migration between vSphere Hosts or Clusters.
(I suspect NFS mounted VMs could be exposed to OpenStack KVM deployments to allow conversions to QCOW format).
Most of the HCI-based products don’t expose their system as a generic storage device so this could be useful for a lot of
migration projects and allow the product features to be adapted to additional “private, public, hybrid cloud storage” use cases
beyond Backup. But making backups simple is a great problem to focus on initially but the software architecture will allow
for the company to track customer funded enhancements easily.
Easy UI – provide vSphere credentials, pick VM’s, set policy and forget it with data rolling out to AWS S3 for retention and retrieval.
The core scaleout, data management and metadata capabilities apply to HyperV, KVM/Xen and Container environments in subsequent
software upgrades by adding additional software as the market demands.
I work for a Hardware VAR and like to evaluate new players. 2 of my smarter co-workers have joined Rubrik early and have sold it to
one of our clients (without our involvement).
NICE TO ADD NEXT: Remote Replication.