It’s unbelievable how amazing the data storage world is, and I have to thank two incredibly compelling startups for reminding me of it. Lately I’ve been worried about the lack of differentiation in the storage startup ecosystem, but these two have proven me wrong!
Well, I have to add that if you look at the container ecosystem, storage startups are popping up like mushrooms but, again, most of them look very similar… on the contrary, NooBaa and Iguaz.io have the most interesting ideas I have seen in a while though and, albeit they are very young (still in a stealth-ish mode with products at a beta-ish level) I really loved what I saw during the briefings and these are the kinds of startups that are really worth listening to find out what they’re doing. If nothing else at least for the idea.
Transparent and smarter storage
They are not competing against each other, but I’d love to talk about them in the same blog because their common denominator lies in the fact that they are building something different from what is currently present in the field.
So, let me start from NooBaa. We are talking about Object Storage here. I know, you may think I’m biased because I love everything about OBS… but it’s not the OBS itself, it’s the approach that makes the difference here.
The idea is quite simple. This is a true SDS solution with storage nodes and intelligence separated from each other. The “brain” is in a VM (or a set of VMs) and it manages the cluster, while the storage nodes (which can be virtual, physical or in the cloud) can be installed everywhere and can take advantage of the storage resources found in the host. The brain has all the info about reliability, performance and availability of each single node and, over time, it learns how to place data correctly to take maximum advantage of the available resources.
But, there’s more. In fact, you can start your cluster by taking advantage of unused resources (and you know better than me how many under-utilized storage resources you have in your datacenter!). And allow me to oversimplify here by saying that hardware TCA of this solution is practically 0 at the beginning of its deployment. That’s great, isn’t it?
But the advantages don’t end here. In fact, when you start to use the product more seriously you can’t build out the infrastructure with the resources you need at that precise moment ranging from physical nodes with plenty of inexpensive SATA-based capacity up to the cloud, if you need its flexibility (as well as manage unexpected capacity needs). And I haven’t mentioned any other feature of the product yet (most of the features are in the realm of what you expect from a modern object store though).
On a totally different wave length, you can find Iguaz.io. They are building a unique storage system that does much more than storing data and it is the perfect match for big data and, more in general, cloud-based analytics applicatgions. In practice, thanks to Iguaz.io, it is possible to build a cloud-like infrastructure where you save your data in any form and you can access it through many different protocols and applications. The concept is quite complex but the idea is that the system can ingest data and perform normalization operations on it, so that you can access it as is, as an object or as a DB table or whatever… concurrently! It’s exciting, powerful and makes it easier to build a data lake as well as the compute infrastructure and applications infrastructure on top of it. It’s like having many of the data services available from AWS (or Azure) without the complexity of managing the infrastructure and copying/converting data… and as far as I can see, it is incredibly fast.
I wrote about smarter storage systems many times in the past, meaning the kind of systems that can do more than just store data safely. This is a concrete example of this class of systems and could really make data lakes much more smarter than what they are today with data available to many different application stacks instantly, when needed.
Closing the Circle
NooBaa and Iguaz.io are doing two different jobs but, in both cases, their approach is really interesting.
Both of them have decided to pre-launch their architecture and follow up with a real product launch later this year… And I think it’s a good choice, there is a lot of innovation here and it takes a while to understand the potential and the implications that their products could have on existing infrastructures.
As I said earlier, I strongly suggest you keep an eye on them (I’ll do and you’ll probably see me mentioning them often in the future). It’s still too soon to tell if they are doing it right but, at least, I’m sure about the fact that they are doing it differently.