What's This?

Congratulations! As your company transitions to software-defined data centers, you’ve been put in charge of storage. Dream assignment, but so many choices. The good news is that by using open source storage software on industry standard hardware, you break the uncomfortable constraints of vendor lock-in. The bad news is there are so many choices of configurations to choose from. Where to start?

In this highly charged, competitive environment, you know that two issues will constrain you: time and budget. In other, very familiar words, do the most you can with the least money as quickly as possible, and deliver an optimum solution.

Your team members say, “Okay, so we’re going to build a storage server farm for use by all departments and (eventually) all workloads.”

You’re perplexed. With all the varying workloads you need to support, which standard storage server is going to suit all of them? Is there one that will meet the needs of long-term archive storage, high-transaction workloads, and big data analytics?

Software-Defined Storage Provides The Solution

Then your team arrives at a great insight. As software-defined storage disaggregates the storage hardware from the storage software, it gives you maximum flexibility in hardware selection. You can use whatever type of industry standard hardware is best suited for the workload you need to support, while using the same storage software to deliver workload-optimized performance.

Members of the team start busily discussing and sketching ideas, while others start shopping for standard hardware and storage software. You say, “STOP!”

You tell them, “Let’s not reinvent the basic wheel here. Let’s leverage available reference architectures to save ourselves time and accelerate our process.”

You point out to the team that reference architectures are written by architects for architects to help speed and improve the development of software-defined data centers for today and tomorrow.

“They’ve already done a lot of benchmarking and tried out all kinds of use cases on all kinds of hardware. That kind of testing would take us months,” you explain. “It has already been done.”

In fact, you tell them, many of the standard storage server manufacturers have participated in the development of reference architectures, providing great insight into how to best deploy their servers for each of your company’s use cases to help you quickly adapt their products to different requirements. You have quantitative data that you can consult on just how different software and hardware combinations will perform.

You point out that some of your workloads will demand high input/output operations per second (IOPS), while others are capacity- or throughput-biased. Using reference architectures allows you to shortcut the testing that would otherwise be required.

Future-Proofed

“Best of all,” you intone, “we won’t be locked into particular appliances or manufacturers. As hardware continues to innovate, we can keep pace by rolling out new hardware platforms, while still using the same storage software layer and operational expertise.”

You then remind your team that this will also afford them the opportunity to give and take from the open source community to leverage a much larger pool of experts. “This helps everyone,” you observe, “and connects us to a massive set of helpful resources contributed by architects like ourselves.”

Red Hat is the curator of reference architectures in the world of OpenStack and cloud-based storage, maintaining libraries of in-depth reference architectures, all free for the taking. Search Red Hat’s publications and data assets here.