Disclaimer: I recently attended Storage Field Day 8. My flights, accommodation and other expenses were paid for by Tech Field Day. There is no requirement for me to blog about any of the content presented and I am not compensated in any way for my time at the event. Some materials presented were discussed under NDA and don’t form part of my blog posts, but could influence future discussions.
For each of the presentations I attended at SFD8, there are a few things I want to include in the post. Firstly, you can see video footage of the Qumulo presentation here. You can also download my raw notes from the presentation here. Finally, here’s a link to the Qumulo website that covers some of what they presented.
Qumulo is a 3.5 year old company from Seattle, WA that officially launched in March 2015. Interestingly, the founders invented OneFS and Isilon scale-out NAS. They pointed out that nowadays managing storage isn’t the big problem, it’s managing the data. Qumulo want to “[b]uild invisible storage that makes data visible”.
Ostensibly, Qumulo’s QSFS is a software-only layer that can work either with Qumulo’s appliances, roll your own hardware, VMs or cloud.
Qumulo sells storage to people that care a great deal about their data
Qumulo stated that their key design goals were to be:
- Hardware agnostic;
- Fast and scalable; and
- File and object aware.
They want to achieve this by building “data-aware, primary scale-out storage systems that provide real-time analytics to give visibility into data usage and performance at scale”. Sounds like big ambitions, but Qumulo seem confident they can pull it off.
The solution is ideally suited to commercial HPC and large-scale unstructured data environments, with real-time analytics that provides the ability to curate and manage the data.
Qumulo have taken a flash-first hybrid design approach with their hardware. In their opinion this maximises both price / performance and price / capacity ratio. I tend to agree with this approach, given the relative sluggishness that we’ve seen with regard to the drop in price of flash.
Qumulo also employ a SaaS-type software delivery model, leveraging a subscription model not perpetual licenses to provide pay-as-you-go access to “continual software innovation”.Everything about the solution is 100% programmable, with a public and self-documenting REST API with interactive API explorer built-in to the product. Interestingly, the development and release cycle is tight, with new iterations of the platform being released every 2 weeks.
[image courtesy of Qumulo]
You can check the specs for yourself here, but think of the QC24 as the fast one and the QC208 as the big, dense one. Of note, Qumulo state that the supported cluster size is 4 – 1000 nodes. I think someone asked them about that and it hadn’t been tested with physical nodes yet. But the argument from Qumulo was that it was technically supported. Connectivity is also slightly different, with the QC24 Hybrid model sporting 2 * 10GbE SFP+ and the QC208 sporting 4 x 40GbE QSFP+. Cache is also slightly different to support the different capacities. Fine, so they’re not really the same.
Closing Thoughts and Further Reading
Qumulo state that they do “100% sales through channel”. I’ll be interested to see how long that model lasts. It’s something I always like to hear, as I work for a VAR, but oftentimes the reality is never quite as expected as the vendor grows and seeks new revenue opportunities.
In any case, as Qumulo pointed out in their presentation, we’re witnessing a shift in enterprise storage, moving from
- Hardware-based to software-based;
- Proprietary operating environments to linux-based OS platforms;
- Hard drives -> flash-first hybrid appliances;
- Petabytes of data to billions of files / objects; and
- Storage management to data management.
Qumulo seem well-positioned, on paper at least, to flourish in this new world of enterprise storage. I like their approach to data rather than storage management. I’m really interested to see how they go over the next 12 months, particularly with the channel-only model and the rapid software development cycles.