The data backup industry has not experienced a radical change within the last 20 years. Many of the same players...
are still big in this industry and they continue to churn out new releases every few years, with small improvements to the offerings. But no one has done anything really innovative for a long time.
The architecture of these types of products also has not changed for quite some time. There are media servers and clunky catalogs that always seem to be part of every offering. No one really likes to do backups, and it seems like the backup vendors know this, but are focused on making it even less enjoyable.
If it were not for backup to disk and virtual machines (VMs), the backup industry would probably be as relevant as Vanilla Ice these days. Backup to disk gave vendors and backup admins something to think and talk about for several years. Heck, they are still debating this topic, trying to figure out if tape is dead yet. Disk allowed organizations to back up and recover faster, without having to retrieve tapes and other challenges that tape can provide.
The virtual machine revolution shook up backup vendors a little. It was clear that customers were not going to want to continue to use in-guest agent backups for virtual machines long term. This new virtualized world would demand a better way to backup these new objects. The idea of creating a generic VM only to have to restore the contents back to it was an ugly situation.
So, VMware and the backup vendors eventually came to the conclusion that backing up the entire VM at the hypervisor level was a better method. This allowed an easier method to restore an entire virtual machine as an object and, eventually, files within a virtual machine.
Ready for change
Given that very little has changed with backups, and organizations have been widely adopting flash and other technologies -- such as hyper-converged offerings that offer simplification of management -- it makes sense that backup is ready for this type of disruption. Hyper-converged and a simple management story are exactly what backup needs.
I have already started to see some movement in this space, but the players are either very new or not ready to publicly talk yet. So, for the scope of this article, I'm going to talk in general terms and about what I think would be cool features to improve the backup space.
There are two major parts of what hyper-converged technology could offer the world of backup. The first is the hardware side that operates as a target for backups. The second would be the backup software that controls and orchestrates the entire process.
On the hardware side of things, organizations have been using specialty appliances for some time now. They are dedicated hardware devices or appliances that are architected just to be good at ingesting large amounts of backup data. They can be pretty expensive and exist only for a singular purpose, which could make hyper-converged technology an attractive alternative.
What if you could take the hyper-converged approach to building out a backup target offering? It would be built on commodity server hardware, with each node contributing CPU, memory and disk capacity. This is the same approach that a bunch of hyper-converged vendors are offering as a platform to put VMs on. This would flip the idea and engineer a scale cluster that is primarily focused on ingesting backup data. The scale-out nodes can all offer capacity and performance to the offering, which is likely to easily beat or compete with today's dedicated appliances.
This approach would also offer an easy scale-out method for capacity and performance to organizations, as their backup needs increase. Replication would also be an important part, along with compression and deduplication. These are all features in today's hyper-converged offerings that would only make sense in a backup product.
The software is the other part that needs more help than the hardware side of things. Backup software is not fun to use and it's either too time-consuming or too complicated. People don't want to spend a lot of time managing their backups; they want to know they are completed and that the data is good. Many of today's backup products take a team to manage if the organization is large, and smaller IT shops cannot afford to dedicate someone to managing backups.
This is where the simplicity of today's hyper-converged offerings could be brought to the backup software space. Think about how easy it is to create and manage a virtual machine in these modern products. What if creating a backup policy and assigning it was equally as easy? It's something that could be done in a simple Web interface with minimal clicks.
Also, when it comes time to restore a file or VM, that should be equally as simple. Admins should not be forced to work with the unfriendly catalogs that exist in today's backup products. This process is ugly and time-consuming. What if you could use a Google-like search to feed in the parameters to locate you file or VM? Think about this; how simple would that be? Enter a search that included the VM name and the file name, and the product returned results within seconds. People might not hate backups as much anymore.
I hope to see some good, new products offering something similar to what I've described here. If you have to manage backups, keep your eyes open to find out when you might be released from your backup shackles.
The hyper-converged hype isn't going away soon
The hyper-converged infrastructure bandwagon picks up speed
Comparing hyper-converged and converged infrastructures
Hyper-converged vendor lock-in: Worry or not worry?