Do you need responsive storage that provides hundreds of thousands of input/output operations per second (IOPS) at sub-millisecond latency? With Qumulo’s v3 software release, you can now get 1.2 million IOPS in a 4-node cluster!

Qumulo v3 hybrid file software: Cluster Activity

Qumulo hybrid file software was designed with several core principles in mind:

1. Hybrid cloud (including multi-cloud) would become the standard enterprise IT model,
2. Data volumes would someday be in the multi-billions of files, and
3. NVMe would be the predominant storage interface.

These three core design principles turned out to be not only insightful back in 2012, but ultimately correct today. NVMe is now becoming democratized in the industry with numerous applications being used that require this extreme low-level of latency.

Qumulo continually updates its file software every two weeks to increase its performance and features, which is automatically made available to all licensed subscribers. Unlike other legacy storage offerings, there is no need to purchase another product or a new software version. Upgrade your Qumulo software and BAM! your applications will also be running faster.

Qumulo also developed the industry’s first all-NVMe file storage platform. Our P-series family, combined with our file software, is how we deliver blazing fast performance – how to get 1.2 million 4k read IOPS in a 4-node cluster. As you scale your Qumulo infrastructure – which is as easy as simply plugging in another node – rebalancing happens automatically – the IOPS scale accordingly. A 100-node P-series cluster has 30 million IOPS!

Another unique Qumulo differentiator is that our software performs even when your storage is 100% utilized. Qumulo’s file storage provides consistent, sustained performance regardless of how full the capacity is. We have no restrictions on the amount of usable storage you can consume to maintain this extremely high performance. Feel free to use what you paid for!

Qumulo Performance Series Platform: P-Series

A bit of background:

Remember that all-NVMe is not the same as all-flash (See my previous blog on “Why Go All-flash when You Can Go All-NVMe?” for a refresher). All-NVMe gives you blazing fast speed because it is directly connected to the PCIe card with 8 direct lanes and 64,000 parallel queues that can each handle 64,000 commands which enables 4 billion parallel commands. When you’re talking with storage array vendors SSD/NVMe performance, make sure you are talking about all-flash with NVMe.

NVMe Specifications and Performance

“IOPS” refers to the maximum number of reads and writes to non-contiguous storage locations. In an HDD, a spinning disk, these non-contiguous storage locations require the drive to continually move its head to the new correct location. These movements are not only incredibly time consuming, they also significantly contribute to the reduced life of the drive. Flash or SSDs all but eliminate this head movement.

With the rapid price drop in SSDs, enterprise organizations are increasingly turning to Qumulo’s NVMe-based platforms not just for high-performance workloads, but also for general file storage and SAN replacements. In recent years, as you can see in the chart below, SSD prices have dropped significantly. With the price delta so small between SSDs and HDDs why would you not take advantage of the latest technology and the extreme low latency it provides?

SSD vs HDD Prices 2015- 2023

When disk drives were slow and cumbersome (i.e., HDDs), legacy storage file software could get away with being less efficient, the bottleneck was the disk drive itself. With all-NVMe and the extreme low latency it provides, the efficiency of the storage file software is now critical. As mentioned above, Qumulo’s v3 software release was designed with all-NVMe in mind; it excels by providing fast, responsive storage that provides hundreds of thousands of IOPS, and supporting high bandwidth connections.

Qumulo’s hybrid cloud file storage, combined with our all-NVMe P-series platform provides organizations with extreme low latency: 1.2 million IOPS in an entry-level cluster. This extreme low-latency is often required to utilize today’s artificial intelligence and machine learning applications to mine your file data for actionable intelligence.

Professionals that can benefit from this extreme low-latency are healthcare professionals needing instantaneous access to your medical imagery. scientist with demanding computational research workloads and artists and producers in video production and editorial workflows.

Don’t get slowed down by your file storage system that wasn’t designed for NVMe. Learn more about Qumulo and contact us for a demo!