Datera is the only storage product that’s built to natively support a declarative, intent-based paradigm for configuration and management.
That’s a mouthful, and it’s a new idea in storage, so I’m going to illustrate it with an analogy to help clarify what it is and why it’s going to revolutionize the way you manage storage.
Let’s dive into the world of manufacturing for a moment.
Imagine that we’re in a traditional, old-school manufacturing paradigm and we’re going to mass produce a new part for a car engine. It’s a custom part with a detailed specification and irregular features. Maybe there’s some very precise shapes and angles, or some holes that need to be drilled in just the right places to just the right depths so the part fits perfectly in every engine.
In a traditional manufacturing model, how is this achieved?
First, schematics are designed for the part, detailing every dimension, measurement, and physical feature. These schematics tell us what we are going to build. But then comes the difficult part.
A tremendous amount of time and energy goes into determining how the part will be built — the schematic must be translated into a start-to-finish, step-by-step manufacturing process in a factory. Entire custom tooling such as molds, presses, etc. may need to be built from scratch for the purpose of achieving just one step of the build process for this part. Then, with all of these resources poured in, eventually a process is in place to have the part manufactured.
This model has some serious drawbacks. Most notably, once the resources have been committed to produce the part, any changes are tremendously expensive.
The entire process is rigid and inflexible. It has been highly customized to this one part and its exact specifications, so we may have to make massive changes to the tooling and infrastructure for very small changes or adjustments to the part. This approach only makes sense when a part can be produced at massive scale, to justify the large upfront investment. Prototyping, experimenting, iterating, or agile adjustments are impossible.
Now let’s consider the advent of 3D printing and how that has changed the traditional model.
In an ideal case, the workflow is greatly simplified. Yes, you still need the plans detailing the physical dimensions of the part — the what. But once you have those schematics, you can simply hand them to the printer and its software will figure out how to build your part.
The irregularity or complexity of the part’s physical features are abstracted away for you; just tell the printer what you want and it will build it for you. If you need to make changes to the part, simply change the schematic template and the printer will adjust the way it builds the part. No lengthy and expensive ramp-up time or custom build-out are necessary. Prototyping, and changes or adjustments, are cheap, fast, and straightforward.
We’ve all heard about how 3D printing is revolutionizing manufacturing — spawning entirely new industries and opening up countless avenues for innovation.
In the storage world, traditional storage management is like old-school manufacturing.
Deploying an application is a two-step process: first figure out the requirements of the application, then a large investment of time and energy go into building a custom scaffolding for it, carefully planning and predicting its capacity and performance needs and managing the hardware on which it is deployed.
You first decide what your application needs, and then you tell your storage infrastructure how to meet those needs. If the needs change, it’s a potentially work-intensive process which could involve anything from small tweaks to ripping out the entire application and moving it to different hardware.
Datera brings the equivalent of the 3D printing revolution to the storage world, transforming storage management and economics.
Our software runs on standard mixed-media x86 servers, providing a declarative, policy-driven management abstraction so that customers can easily specify needs of their application and the value of the data, while the software determines the best placement and configuration for that data. As the value of data changes over time, customers can easily move workload transparently across different storage platforms through simple policy changes.
Now, deploying an application is a single step — you tell our software what the needs of your workload are and we figure out how to deploy it on our infrastructure based on those needs. Changes are seamless and transparent to the workload.
Does one of your workloads suddenly need more performance?
All it takes on your end is a small change through our GUI or API to bump up its QoS parameters, and based on that, the Datera system will transparently adapt to you and the application. Perhaps nothing will change, if the existing deployment configuration meets your new needs. Perhaps parts, or all, of your application data will be moved from disk to flash, or will move from a hybrid storage node to an all-flash storage node. You don’t have to worry about the details — you just see a sudden boost in the performance of your application.
If the performance you require cannot be met with your current hardware, our software will let you know, and expanding capacity or performance is as simple as adding new nodes to the system. Again, Datera will dynamically and seamlessly begin taking advantage of the new hardware to fulfill application requirements as soon as it is added, without any intervention.
This way of managing storage is not just a superficial veneer implemented by the GUI or REST API. It is only possible because our control plane and data plane were built from the ground up to support this way of managing resources and deploying workloads, building abstract models of the storage media and other hardware infrastructure, and policy-based management of those models, so that workloads can be managed simply and consistently from a single locus of control.