Forward Thinking Backups

So I thought I was the only one that was terrible at backups. Even after losing all the data from my first PC, a sweet Packard Bell running a hot 200Mhz Pentium MMX, I never learned my lesson. Maybe it’s just my devil-may-care attitude. Maybe I just like the exhilaration of knowing everything I hold digitally valuable could be gone in a flash. Or maybe I’m just lazy…

Even when I do get around to backing up, I know I’m doing it wrong. Is there any worse feeling? I have things on Dropbox, Google Drive, I think a few Word docs on an encrypted on an old flash drive, and maybe a 2 year old Time Machine backup on a drive that I’m not even sure powers on. Hardly a robust solution, my data recovery time has the latency of remembering where the heck I last saved something. Luckily, I’m not in charge of any enterprise deployment. Here’s the problem, most backups stink there too.

Now there are a hole slew of different reasons why backups suck in the enterprise. Those deployments are less concerned about making sure you don’t lose the complete series of Boy Meets World and more worried about dedupe, replication, indexing metadata, you know…business. I saw a presentation from Rubrik at Tech Field Day that outlined why the situation is a mess. Basically, the backup world has a bunch of moving parts that aren’t great about talking to each other. Between backup software, servers, proxies, catalog databases, tape archives, and offsite storage there are a lot of places for things to go wrong, or get siloed off. Rubrik thinks they have a more complete offering that brings together all of these elements into one manageable service.

Rubrik bases their hardware in what they call “Briks” (which my spell check just loves guys…thanks). These Briks come in a couple of configurations, depending on capacity and capability needed within your deployment. I know, a rack mounted backup storage device. How does this solve anything? Well Rubrik has a lot of stuff baked in their devices to deliver what they promise is converged backup and replication software with globally deduplicated storage. This is powered by their third generation underlying operating system Firefly. Based on the demos I saw, it looks like a pretty mature release. The release is mature, but don’t worry the Rubrik team peppered plenty of Joss Whedon references throughout.

With Firefly, Rubrik thinks they can remove a big part of the legacy backup architecture. With a Brik in the rack, they can removed the chain of servers and proxies used for backup and search, as well as the need for tape and offsite archiving. They do this by provides physical horizontal scale out with additional Brik appliances, as well as with robust support for S3-compatible cloud storage. Most impressively of all, this isn’t simply localized to one datacenter. Rubrik demonstrated the ability to have their Brik appliances running at different locations, with VM’s all being fed from remote offices. This could still all be managed within their cloud interface, and from their demos, this seemed to be fairly intuitive, although I’m admittedly not much of an expert. Best of all, it what is fast becoming a trend in the enterprise backup world, they’re offering fast Google like search options on those backups. I’ve seen this from a couple of vendors, but when dealing with the volume of enterprise backups, it’s still impressive to see it work seamlessly, although I feel in a few years this will be table stakes.

All of these tools are based on some impressive low level engineering. Rubrik touted their proprietary file system, Atlas, as the key to making their distributed backups work. Atlas was designed from the ground up to handle those kind of workloads with a Global Namespace. This is complimented by a separate Distributed Metadata Store called Callisto, which I believe is responsible for their fast handling of search across the backups. What Rubrik hit on over and over again was that this was designed with scale in mind.

Overall, if you find the current enterprise backup situation to be as fragmented and error prone as my personal situation, you might want to give Rubrik a look. Their solution seems laser focused on backups specifically, allowing for unified management no matter where they are located. They also provide for scale out, either with additional appliances, or into the cloud as need dictates. I’ve seen other vendors offering solutions based on the larger idea of managing “secondary storage” (looking at you Cohesity). Rubrik’s solution could certainly be applied to that broader palate, but I liked their built from the ground up focus on backups.