Genomics England

What its that RAID its we inherent replacement love about architecture the technology, activeScale is underpinned the system intelligent is by dynamic placementoferasure-coded data.

,

David Ardely
Head of Technical Delivery, Genomic England

SOLUTION OVERVIEW

KEY BENEFITS

Genomics England has ambitious aims. The organization was created to sequence the genomes of an enormous number of people, generating new insights that can help improve treatments—while also accelerating the development of the UK genomics industry.
Within a few years of its inauguration, the organization proved that it could reach its initial goal of sequencing 100,000 genomes. And it was rewarded in 2018 with a new goal of sequencing up to five million genomes over five years.
Unfortunately, the existing network-attached storage (NAS) solution used for storing genomic data was not up for the task. The NAS, which was storing 21 PB of data, could not be scaled further—it had reached its node-scaling limit.
As the NAS approached capacity, performance suffered. Meanwhile, the organization could not ensure resilience for all this genomic information. Though data was backed up to tape, Genomics England lacked a robust disaster recovery strategy that could restore data rapidly in the event of a disaster or another significant disruption.
The organization needed a solution that could store and protect fast-growing data volumes. "We needed something that's much more scalable than existing NAS solutions—an infrastructure that could grow to hundreds of petabytes," says David Ardley, head of technical delivery at Genomics England. A new solution also had to facilitate simple, flexible access to data by more than 3,000 researchers around the world.

USING QUANTUM ACTIVESCALE OBJECT STORAGE

Genomics England consulted with Nephos Technologies, an independent UK-based data services organization to design and implement a new storage solution. Together, teams from Nephos and Genomics England reviewed current requirements, projected future needs, and then designed and implemented a new, multi-faceted solution.
The storage solution incorporates a WekaIO high- performance file system, Mellanox high-speed networking, and Quantum ActiveScale object storage. It creates a two-tier architecture that combines flash storage plus ActiveScale object storage, which serves as a long-term data lake repository. The initial deployment included 1.4 PB of flash and 40 PB of object storage. The two storage tiers—each of which can be scaled independently—present as a single hybrid storage environment. As a result, researchers have the flexibility to query data in a highly randomized fashion.

TAKING ON NEW CHALLENGES DURING THE COVID-19 PANDEMIC

Within a few years of deploying the new storage environment, Genomics England needed to expand again. The emergence of the COVID-19 pandemic in early 2020 presented new, urgent challenges for the global medical-scientific community, and Genomics England was in a prime position to help better understand who is susceptible to the virus. The organization committed to sequencing the genomes of up to 20,000 intensive care patients with COVID-19 plus up to 15,000 people with the virus who are experiencing only mild symptoms. In addition,
the organization launched a next-generation genomic research platform to enable researchers and drug developers to access and use genomic data in their COVID-19 work.
Around the same time that Genomics England was ramping up participation in COVID-19
research, the ActiveScale solution platform was acquired by Quantum. A Quantum team facilitated a smooth transition for Genomics England.
With full confidence in Quantum and the ActiveScale system, Genomics England moved forward with expanding its object storage environment to accommodate COVID-19 work. The organization scaled to more than 100 PB of capacity.
Scaling was seamless with RAID (Redundant Array of Independent Disks). "What we love about the ActiveScale system is that its inherent architecture is underpinned by its RAID replacement technology, the intelligent, dynamic placement of erasure-coded data," says Ardley. That placement of data eliminates the need for system rebalancing, which can compromise performance and availability.