06/12/2018 |

The Relationship Between Software-Defined, Hardware Agnostic, TCO, and ROI

Following up on our series of articles about the Zero-Waste OpenIO SDS Cluster, I'd like to give you more details about the benefits of being fully hardware agnostic, what this means for the TCO (total cost of ownership) of your infrastructure, and how it helps you get a better ROI (return on investment).

OpenIO's hardware philosophy

How to size an OpenIO cluster on x86 servers

The SDS project started more than 10 years ago. Our first production environment was 1PB, and that cluster grew up to 10PB and billions of objects under management in a few years. The final cluster was not only totally different in size, but we had also to manage different vendors as well as hardware generations.

We developed a technology to take into account how every single resource in the cluster was performing to make the best of what is available: it is called Conscience. I won't go into details about this, we have several articles that can tell you more about Conscience, but this is what powers OpenIO SDS's flexibility and its ability to take advantage of heterogenous hardware in the same cluster.

There are two other important reasons why OpenIO is different. The first comes from the fact that OpenIO SDS is very lightweight. In fact, the smallest installation we are able to perform is on a Raspberry Pi Zero (1 ARM core and 400MB RAM). And this leads to the other important aspect of OpenIO SDS, which is speed. OpenIO SDS’s performance is better than that of the competition because the high level of optimization of our code, and because we can take advantage of all the hardware resources in the cluster. 

Being flexible, fast, and smart is our philosophy, and it reflects in everything we do, including hardware optimization.

How OpenIO SDS helps improve TCO

Planning a new infrastructure is fairly easy, but making it evolve efficiently over time is a different story. 

There are two ways to plan a new storage infrastructure with OpenIO, depending on your initial budget and your storage project. As we have discussed many times, you can start with brand new hardware or recycle existing servers. Beginning with new hardware is always a good choice if your budget allows for it, but there are several situations for which you may want to consider a softer approach and a more limited investment.

Several OpenIO customers started with the latter option and took advantage of new hardware when the project became more important or outgrew their initial capacity expectations.

Having the ability to reuse components of your existing infrastructure is very powerful, but it’s even better if you can mix old and new generations of hardware without compromising on scalability or performance.

Thanks to Conscience you can add as many nodes as you need, and they can each be of different capacity, CPU power, and network connections. They will seamlessly contribute to the total amount of resources available according to their abilities.

This improves your TCO by responding to business requests promptly and without large investments such as fixed configuration expansions, and prevents wasted resources on systems that cannot take advantage of unbalanced hardware clusters.

The ROI impact of OpenIO SDS

The first thing to consider is the kind of investment you need to make to start out with OpenIO SDS. If you start with 100TB usable and recycled hardware, it is quite easy to have a quick return from this kind of investment.

But the real advantage doesn't come from the low entry point; what makes the difference is the granularity of expansions. To expand an OpenIO SDS cluster you can theoretically add a single disk and take advantage of it, immediately. Even though most people think of object storage as static environments for long-term archiving, our customers are now buying OpenIO SDS because it is faster than its competitors, and they run workloads that a few years ago were unthinkable on object stores.

Performance comes from flexibility, and a better ROI comes from the fact that things evolve quickly. You may choose an object store for long-term archiving today, but want it to be ready for other types of workloads when you need them in the future. OpenIO can do that, just by adding new hardware that is adequate for the new workload. This is why we work so hard to optimize performance.

Key Takeaways

OpenIO SDS is flexible, smart, and fast, even when taking advantage of different types of nodes in the same cluster. Our customers love how OpenIO SDS works with different hardware nodes in the same configuration and how this help them save money every day.

How to size an OpenIO cluster on x86 servers