When I recently wrote about dynamic data protection, I decided to avoid commenting about other characteristics of traditional object storage; I wanted to avoid what could be considered FUD. But I can't stop myself, and today I'd like just to point out one OpenIO SDS feature that, for me, is much more critical than many others. The ability to run on heterogeneous hardware and start small, very small.
The benefit is crystal clear: with a very low initial investment, businesses are free to grow their storage platforms as needed.
You know we can run on a Raspberry Pi, right? Well, this is probably too small to deploy an object store in production, but it gives you the idea of what we can do.
Not everybody was born with petabyte-sized pockets. The vast majority of businesses start with a single application and a few terabytes. Even a local ISP, interested in selling object-based cloud storage services, wants to start small and see the market reaction before making large investments.
Last week, when I was talking at VMUGIT in Lecce (Italy), I met a local ISP. He has an old blade chassis with 10 blades. These blades are end of life, no longer supported by any virtualization platform, and the renewal of their support contract is too expensive. But the chassis is fully redundant, and with plenty of network connectivity, and he can add two 2TB 2.5" drives in each blade. That's huge!
We talked about potential applications for object storage, including backup, cloud storage for his customers, OIO-FS for local NFS/SMB connectivity, and so on. Eventually I suggested that he start thinking about building a first OpenIO SDS installation with the hardware he already has, the blades.
When small really means small
The minimal OpenIO SDS configuration in production is three nodes, one disk per node. And the nodes could be different from each other. But ten is a great number to start with, enabling erasure coding from day one for better efficiency.
Ten blades (2*2TB) is 40TB raw, which corresponds to more than 26TB usable with EC6+3. And you don't need any hardware support contract for it. This configuration can lose up to three blades and still function. This is probably better resiliency and availability than any other traditional storage device in your infrastructure today.
What about cost?
OpenIO SDS is open-source software. You can install and run it for free. This is a good option if you want to do a PoC or test your applications. We can also provide professional services to help with initial deployment and tuning. Then, when everything is set up, a standard support subscription can be added.
So, how much does it cost to start with OpenIO SDS? It could be €0.
What happens next?
No matter what happens in the future, this configuration will be fully upgradable with different nodes. Not just a different form factor, but also CPU, network, RAM, and disk capacity. Conscience technology will take care of managing it correctly to get the best out of all the resources available in the cluster.
And no matter if the configuration grows up to petabytes or multiple datacenters with a geo-distributed configuration. OpenIO SDS’s support subscription doesn't change, we only charge for usable capacity.
OpenIO SDS can start really small. It's the object store for the rest of us.
Most traditional object stores need fixed configurations, a minimum number of similar nodes, large hard drives, and more. Not us.
OpenIO SDS’s lightweight and unique design makes this possible. It really is flexible, hardware agnostic, and future proof. Try it out for yourself and see.