Insights, Best Practices and Forward Thinking from the Customer Facing Team, Solution Architects and Leaders on Extreme Performance Applications, Infrastructure, Storage and the Real-World Impact Possible

The Role of Features in Flash Storage

by Tim Stoakes on February 4, 2015

In this episode of the Architecture Matters Blog Series, I’ll be covering the rising importance of data management features for all-flash arrays.

Let’s start first with looking at why data management features are critical for an all-flash array: our first generation product was addressing the disk performance gap for tier-0 storage. In those days, what mattered most was best performance/price ratio, and with relative small deployments, where the importance of data management features was not critical.

Today, all-flash arrays can now beat the cost of disk.

As customers consider flash storage as their primary storage platform, they demand features that are comparable to their legacy disk solutions. After all, they want data and array management to be as easy and complete as before.

So how does any all-flash array get features comparable to legacy disk solutions?

One obvious way would be to take a legacy disk array, and replace all its drives with SSDs and take advantage of existing operating system and the features.

The challenge with this approach is that the disk array’s operating system was engineered for disk and its latency that is obsoleted when replaced by flash. The use of SSDs does not unlock all the potential that flash has to offer both on cost and performance, so the end result is still really a slow and expensive disk array.

The next logical step for the vendors who have a ground-up designed all-flash array is how to provide data management features.

One approach is using separate heterogeneous storage virtualization products creating a two box solution. This approach increases unnecessary cost due to separate hardware needed for the storage software controllers, as well as two management interfaces making the solution complex and expensive.

The better approach is building a flash platform that is powered by an operating system that is built from the ground up to run on an optimized design all-flash array. It is managed by a single simple and intuitive graphical management interface, one that not only manages a single array but all arrays (single pane of glass) to simplify management to the maximum extent possible. Not surprisingly, this is the Violin approach.

At Violin, we look at both developing our own software as well as purchasing additional technology. We call this the hybrid build + buy strategy, and with it we have been able to integrating the best of both worlds into our single operating system.

The combination of getting in technology that is field proven, as well as developing and innovating on top of that with one of the best storage development teams in the industry is how we accomplish feature comparability in primary storage.

Today, Violin all-flash arrays boast the most advanced flash operating system in the industry, not only offering the “basic” features such as app consistent snapshots, thin provisioning, clones and asynchronous replication over IP, but the ability to do things such as continuous data protection (my favorite feature), mirroring and stretch cluster for the highest availability architectures possible is truly unique. All of this is then topped off with RESTful APIs, OpenStack and VMware integration and the most simple and elegant management solution in the industry called Symphony.

Want to know more? Contact us today, and we would love to talk you through our portfolio and share with you the success stories of other customers that have been using Violin’s software in tier-1 primary storage.

In the next installment of this series, I will share my thoughts around the performance of an all-flash array, a topic that is dear to our heart. Stay tuned!