Over 1PB in capacity. Delivering 90GB/sec bandwidth and 20M IOPS, at 40 µsec latency. Ask your storage vendor one ‘little’ question – “what’s your performance density?”
From web-scale cloud-native workloads to legacy enterprise applications, Pavilion seamlessly handles your exponentially growing requirements today…and into tomorrow.
The industry’s first Hyperparallel Flash Array is proven in the world’s largest NVMe-oF deployments with the most demanding customers who are making the impossible possible.
Designed for the Next Wave
Pavilion is the inventor of the world’s first Hyperparallel Flash Array. Built for NVMe from the ground up, we target the highest performance big and fast data workloads to speed data analytics and derive actionable insights.
With linear scaling, you can unify legacy enterprise and modern workloads at a reasonable cost without the need to restructure your private or hybrid cloud.
A single fault-tolerant array provides 90GB/sec bandwidth and 20M IOPS, at 40 µsec latency. Our HFA packs 1+PB with RAID-6 protection, 20 storage controllers, 40 @ 100Gb Ethernet or InfiniBand ports into 4RU.
Trusted by leading brands and backed by Tier-1 investors, Pavilion is nimble like a startup, yet a veteran of the world’s largest NVMe-oF deployments.
Standard Ethernet networking, standard NVMe SSDs, standard NVMe-oF drivers. No esoteric ASICs or FPGAs. No host software. Free host resources for scaling applications, not storage. Impossible? Check it out!
A fault-tolerant architecture supports up to 20 storage controllers and 40 @ 100Gb Ethernet or InfiniBand ports. Disaggregate storage from compute, compose logical volumes on demand, without the performance bottlenecks of legacy dual-controller architectures. Patented technology delivers 20 million IOPS and 90GB/sec throughput in a 4RU modular chassis holding 1+ PB of standard 2.5” NVMe SSDs.
The World’s first Hyperparallel Flash Array hardware is only part of the story. Our operating system is based on the NVMe protocol, and purpose-built for modern, latency-sensitive applications. Security, reliability and management features of SAN, and the speed and latency of direct-attached storage.
Change the way you buy storage. Redeploy existing NVMe SSDs into the array, acquire the latest drives directly from the OEM. With four separate storage zones, mix NVMe drives for performance, endurance, capacity, and cost. No fork-lift upgrades; no delays for drive qualification; and no markups!
Online reservation systems collect, move, store and share information from a wide number of databases across
the globe. Delivering the right prices, for the right flight or hotel at the right time, in real-time requires a massive
amount of parallel processing. Real-time analytics are critical, as is having enough capacity to store and compare
against a large inventory of customer options.
See how Pavilion Data transformed a 120 node, 1.4PB parallel processing cluster using Apache Hadoop® and Spark® from server-side storage with extremely poor utilization to an NVMe-oF rack-scale flash architecture with a single pair of 4U Pavilion appliances.
For one of the world’s largest financial institutions, backup and compliance of a Greenplum data lake are of paramount concern. With a centralized cluster that has grown to more than 1PB over a short period of time and a strict backup window, a new approach to addressing high-speed backup was required.
The DevOps team of a leading consumer online services provider was unable to meet their charter of rapidly deploying containerized workloads. See how they created a “Containers-as-a-Service” model by migrating off server-side storage to Pavilion’s Hyperparallel NVMe-oF storage solution.
The ability to ingest, analyze and make decisions quickly on vast amounts of data is the lifeblood of today’s organizations. Whether driven by regulatory requirements, gaining competitive advantages or increasing operational efficiency, you need speed and accuracy to be a game-changer. At Pavilion, we empower our customers to make better decisions, faster.