March 22, 2014

Qumulo: An Enterprise Storage Startup Focused on Scale, Simplicity, and Efficiency

Enterprise storage is perhaps the most innovative area of IT these days, with exciting startups springing up right and left. Today, that scene welcomes Qumulo, who are building a new storage platform focused on scalability, efficiency, and simplicity. Qumulo catches my eye for two reasons: The team is heavy with Isilon experience, and CTO Aaron Passey really impressed me with his work at Clustrix.

Rethinking Storage (Again)

I am a revolutionary. I want the storage industry to radically transform. I want to throw out the old baggage of SCSI and RAID and rethink storage from the ground up.

But I am also a pragmatist. I see the inertia inherent especially to storage. I know it will take decades to move away from monolithic SCSI-centric arrays.

The solution to this quandary lies in the world of startups. When planning Storage Field Day 2, I focused on “new thinking”, and I launched the Next-Generation Storage Symposium with that same thought. I believe that, by highlighting innovation and radical options we can exert pressure on this slow-moving industry. Eventually, things must change; we can hasten that transformation by moving quickly to adopt new ideas.

One of the greatest challenges for storage is scale: It’s incredibly difficult to build a system that grows without becoming unbalanced. This difficulty is compounded by legacy protocols and applications that demand a “fake disk” instead of using a more-adaptable access mechanism. It’s not enough for an array to double in size; my dream storage system would be able to scale up and down by orders of magnitude!

This explains my fascination with application-specific storage (Tintrì, Clustrix, SolidFire), object storage (Amplidata, Scality, Caringo, HCAP), and cloud (Amazon S3, Panzura, Nasuni). These systems suggested a way forward that would allow storage systems to do what they do without “general-purpose” baggage. But not all will succeed!

Why I’m Interested in Qumulo

We don’t know much at all about what Qumulo is doing. Until today’s announcement, they have been very stealthy indeed! But we do have two facts to work with: Their team and their stated goals.

Let’s start with what they say they are doing. Their “About” page has a single paragraph (emphasis is theirs):

Qumulo is a Seattle-based enterprise data storage startup. We believe that storage should be simple, scalable, and efficient. Formed in 2012 by leaders from Isilon Systems, Adobe, and Wily Technology, Qumulo is reimagining enterprise storage.

The fact that scalability is a stated goal is not surprising. After all, that’s what Isilon was all about! But it’s great to see that work continuing! But it’s almost impossible to add scalability after the fact. Systems are either designed to be scalable or they are monoliths. No exceptions.

Efficiency is a no-brainer goal. Everyone in the world is integrating data deduplication, compression, thin provisioning, and tiered storage into their storage solutions. These guys would be stupid not to have efficiency as one of their pillars, and they’re not stupid.

The simplicity angle really gets my interest, however. I met with Qumulo’s CTO Aaron Passey last year while he was working at Clustrix. He impressed me as a person and with his vision for a converged scale-out database solution with integrated storage. That this same guy is putting his efforts into a more general-purpose enterprise storage system is exciting!

Let’s throw some conjecture into the mix: This thing will scale, perhaps over InfiniBand, over many industry-standard hardware nodes. I’m guessing they’re building a hybrid storage platform with flash and spinning disk and that it will include “capacity optimization” as a core element. I bet it’ll be application-focused (“big data” and virtualization) if not application-specific. And I’m hoping it uses NFS for connectivity (or even SMB3!)

Stephen’s Stance

I’m not a “news breaker” but I get excited when I see something interesting come up. Qumulo is interesting, with a solid team and the right vision for enterprise storage. And some serious funding doesn’t hurt! The enterprise storage industry needs innovation, and companies like this are delivering it. And I’m doing my part by bringing attention to them.

Disclaimer: I know Charles Curran of Valhalla Partners and Peter Bell of Highland Capital, both of whom participated in this funding round for Qumulo and are likely board members. I’m covering this because this company interests me. Although I haven’t talked with the team yet, I imagine my phone will ring very soon…

  • Larry Freeman

    Nice recap Stephen. I find the startups like Qumulo interesting as well (cool name BTW). When you describe your dream storage system that can scale up and down in orders of magnitude without becoming unbalanced, sounds like you have the same dream we have at NetApp using system clustering and virtual storage tiering. Throw in 100% nondisruptive operation and you’ve pretty much described our vision.

    I don’t share your opinion, however, of application-specific storage. When you start attaching storage based on specific application attributes, you are back in old silo’ed storage days that have proven inefficient at scale. General purpose storage of the future needs to be agile enough to adapt to any workload, whether its SAN or NAS and either performance tuned or capacity optimized.

    Also, I agree that storage efficiency features like dedupe should be a no-brainer, but not everyone in the world is adopting them, notably Isilon and HDS seem to have missed the dedupe bus. I summarized the current state of dedupe in this blog:–2012


    Larry Freeman
    NetApp Senior Technologist

  • Jerome Lecat

    We welcome Qumulo in our competitive space, to have such a quality team launch a new start-up on concepts that we have been professing for several years is a great validation.

    Stephen, I fully agree with your post. Storage has to change to respond to the “digitization of human life”, but it has to change without changing… I mean without changing its interfaces…

    Jérôme Lecat, CEO of Scality