Qumulo – Storage Your Way

Disclaimer: I recently attended Storage Field Day 20.  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.

Qumulo recently presented at Storage Field Day 20. You can see videos of the presentation here, and download my rough notes from here.


Solving Problems The Qumulo Way

Extreme, Efficient Scalability

Legacy problem

  • Inefficient file system architectures only use 60-80% of purchased capacity
  • Limited capacity and file count scalability
  • Adding capacity and performance complex and often required downtime

Qumulo Solution

  • Get 100% out of your investment
  • Scale to hundreds of billions of files
  • Expand capacity and performance seamlessly

No More Tiers

Legacy problem

  • Data not available in fastest storage when users and applications request it
  • Tiering jobs were slow to execute and often didn’t complete, leaving risk of filling the performance tier

Qumulo Solution

  • Single-tier solution with predictive caching ensures performance SLAs are met
  • Simplified administration and full confidence in access to data


Legacy problem

  • Troubleshooting complexity, unforeseen growth expenditures, and usage bottlenecks plague IT
  • Impossible to track who is using data and for which projects

Qumulo Solution

  • Gain graphical, real-time visibility into user performance, usage trends and performance bottlenecks
  • Proactively plan for future requirements


Flash! Ah-ah

The development of an all-NVMe solution also means Qumulo can do even more to address legacy storage problems.

Intelligent Caching

  • Enables fast random reads from limited and expensive media (SSD)
  • Identifies read I/O patterns and promotes data from disk to SSD
  • Cache eviction policy leverages heat-based strategy

Predictive Prefetch

  • Enables very fast streaming reads
  • Proactively moves data to RAM by anticipating files that will likely be read in large, parallel batches or within a file
  • Constantly automatically adjusting if data is not used at per client granularity

All Writes Go To Flash

  • No special hardware components required
  • Data de-staged to optimise for HDD performance
  • Low latency NVMe makes writes much faster


Thoughts and Further Reading

We’ve been talking about the end of disk-based storage systems for some time now. But there still seems to be an awful lot of spinning drives being used around the globe to power a variety of different workloads. Hybrid storage still has a place in storage world, particularly when you need “good enough” performance in price-sensitive environments. What All-Flash does, however, is provide the ability to deliver some very, very good performance for those applications that need it. Doing high resolution video editing or need to render those files at high speed? An all-NVMe solution is likely what you’re after. But if you just need a bunch of capacity to store video surveillance files, or archives, then a hybrid solution will quite likely meet your requirements. The key to the Qumulo solution is that it can do both of those things whilst using a bunch of software smarts to help you get your unstructured data under control. It’s not just about throwing a new storage protocol at the problem and hoping things will run better / faster though. It’s also important to understand how the solution can scale out, and what kind of visibility you get with said solution. These are two critical aspects of storage solutions used in media production environments, particularly when being able to squeeze every last bit of performance out of the system is a key requirement, and you might be in a position where you need to throw a bunch of workload at the system in a hurry.

Qumulo strikes me as being a super popular solution for video editing, broadcast, production, and so forth. This is one of my favourite market segments, if only because the variety of workloads and solutions that cater to those workloads is pretty insane. That said, when you dig into other market segments, such as Artificial Intelligence and analytics workloads, you’ll also notice that unstructured data access is a common requirement. The delivery of an all-NVMe solution helps Qumulo provide the resources required to satisfy those high-performance applications. But the cool thing isn’t just the performance, or even the ability to scale. It’s the visibility you can get into the platform to work out what’s going on. Managing petabytes of unstructured data is a daunting task at the best of times, so it’s nice to see a company paying attention to making both the end user and the storage administrator happy. I’m the first to admit I haven’t been paying as much attention to Qumulo as I should have, but I will be doing so from now on. For another perspective, check out Ray Lucchesi’s article on Qumulo here.