SAN DIEGO, Calif. – A short walk around the ScaleMatrix campus provides a glimpse into the full spectrum of possibilities in genomics and life science. Within the ScaleMatrix data center, high-density cabinets help power the IT operations of billionaire J. Craig Venter, the life sciences pioneer who led the effort to sequence the human genome. These include the J. Craig Venter Institute and several companies founded by Venter to apply machine learning techniques to genetics research and create better biofuels.
In a building next door, the next generation of genomics entrepreneurs are setting up shop at the ScaleMatrix Life Science and Technology Launch Center, a startup accelerator operated by ScaleMatrix and one of its customers, Diagnomics. The spacious digs include the usual co-working offices, kitchen area and foosball table. But there’s also a room you don’t see at most startup hubs – a human genomics laboratory outfitted with gene sequencing equipment.
The accelerator is optimized for new companies launched by grads of the University of California at San Diego (UCSD), one of the nation’s leading research hubs for genomics and life sciences. In addition to the genomics lab, the startups have access to high-density computing horsepower from the ScaleMatrix data center, as well as low-latency network access to key databases at the San Diego Supercomputer Center.
The Opportunity in Specialized Computing
Amid the growing dominance of giant cloud computing platforms, ScaleMatrix illustrates the opportunity in data centers that focus on a specialized computing niche. ScaleMatrix is a San Diego-based colocation and cloud company built atop a custom cabinet that supports loads of 52kW per rack and beyond.
This high-density hosting has proven to be a good fit with San Diego’s active cluster of companies and universities focused on the life sciences, much of which is enabled by high-performance computing (HPC) and data analytics.
“These technologies are pushing the envelope,” said Chris Orlando, the Chief Sales and Marketing Officer and a co-founder of ScaleMatrix. “We’ve had to morph our business to address these customers. In 2011, we said ‘we can handle 52kW.’ People said ‘who cares? That’s a neat toy, but nobody needs it.’
“In 2016, that changed dramatically,” Orlando added. “We have people from around the country seeking us out because they have dense platforms that are pushing the limits of what their data centers can handle. With densities and workloads changing rapidly, it’s hard to see the future.”
The Cabinet As A Data Center
The key building block for ScaleMatrix is the Dynamic Density Control (DDC) cabinet, which was designed for efficiency but also supports extreme density. The DDC is an extra-wide enclosure with two compartments: a bottom compartment housing a four-post rack (with extra depth to accommodate HPC servers), and a top compartment containing a heat exchanger.
The six-foot deep cabinet is sealed, with air filters, humidity control and fire suppression handled within the enclosure. Air is recirculated within the cabinet, with cool air delivered to server inlets via an 8-inch air plenum in the front of the rack. When exhaust heat exits the back of the equipment, it rises to the cooling compartment and is cooled and recirculated.
A row of Dynamic Density Control cabinets inside the ScaleMatrix data center in San Diego. The cabinets can support up to 52kW of power density, as well as armored cabling connections to protect network and water lines. (Photo: Rich Miller)
Cooling is provided by a fin-and-tube heat exchanger, which is fed by a cool water loop that runs overhead and attaches to the top of the cabinet. The use of containment allows ScaleMatrix to maintain temperatures within a 2 degree temperature range throughout the rack. The temperature within the rack is closely monitored and managed by an in-house DCIM (data center infrastructure management) system, which can regulate the temperature by adjusting a variable control valve on the cooling loop.
In addition to data-crunching for genomics, ScaleMatrix has a group of customers specializing in artificial intelligence (AI) and machine learning. The DDC design has proven ideal for supporting AI workloads requiring graphics processing units (GPUs), which can generate unusual levels of heat.
“This new hardware will be more powerful and energy-intensive than ever before,” said Orlando. “People are looking to support these additional densities.”
One of those customers is Cirrascale, a cloud service optimized for machine learning and GPUS. Cirrascale, the successor to the Verari HPC business, operates several rows of cabinets at ScaleMatrix, which house between 11 and 14 GPU servers per cabinet, including some connecting eight NVIDIA GPUs using PCIe – a configuration also seen in Facebook’s Big Sur AI appliance and the NVIDIA DGX-1 “supercomputer in a box.”
A Tonic for Aging Data Centers?
While ScaleMatrix is using the DDC cabinet as the cornerstone of its high-density hosting business, it is also licensing its cabinet technology to enterprises and rack manufacturers. Orlando believes the ScaleMatrix cabinet design could become an important tool in extending the life of aging data centers, averting the need to renovate entire data halls.
“Stranding of assets is a concern for many data center operators,” Orlando said. “How do you modernize a legacy facility without breaking the bank? It’s not simple, and it can be expensive.
“Our solution is to take a section of that data center and install a cabinet-level solution,” he continued. “You’re vastly improving the capabilities of the resources you’ve already got. Plus, you can now have a 50kW cabinet live next to a 3kW cabinet.”
When the Edge is an Aircraft Carrier
This type of compartmentalized “micro data center” design is also getting fresh attention with the growth of edge computing.
A ScaleMatrix DDC cabinet adapted for use in mobile deployments, with a chiller unit added that can slide out for maintenance. The unit can support 20kW of IT load. (Photo: Rich Miller)
As data moves to the edge of the network, data centers are being right-sized to fit the demands of these new markets, deploying space in digestible chunks.
That’s why ScaleMatrix has adapted its design for remote computing, creating an enclosure with a rear-mounted chiller that can slide out for maintenance. The edge design can support workloads of up to 20 kW per rack.
ScaleMatrix used three of these cabinets to create a data center at the U.S.S. Midway Museum, housed inside the massive aircraft carrier docked at San Diego Harbor. The company was able to complete the installation in just 30 days. The units are also deployed in cruise ships operated by Carnival, which are increasingly reliant on technology.
“We see demand growing around these modular systems,” said Orlando.
Supporting Growth in San Diego Genomics Cluster
ScaleMatrix has nearly filled the ground floor of its data center, and has a second floor available for expansion capacity. Orlando is hoping that some of the company’s future customers will come from the Launch Center accelerator next door.
The Launch Center is equipped with a Tecan Freedom EVO automated liquid handling and pipetting system. (Photo: Rich Miller)
Startups can lease work space at the accelerator at rates ranging from $200 to $3,200 per month, depending on the size of the office. For an additional fee, companies get access to the genomics lab, which is certified by the Clinical Laboratory Improvement Amendments (CLIA). The facility includes lab benches and workstations, as well as freezers, autoclaves, centrifuges, microscopes, biosafety cabinets, and an Illumina gene sequencer. The lab is supported by Diagnomics, a ScaleMatrix customer that offers gene sequencing in a secure cloud environment.
It’s an unusual investment for a data center company. Although many data centers have boosted the amount of office space available at their facilities, ScaleMatrix is unique in its development of an accelerator focused on supporting a specialized business sector like life sciences.
Orlando says it’s the right thing to do. And if the next J. Craig Venter gets their start at the Launch Center, they won’t have to walk far to build out their IT infrastructure.
To view the original article, please visit: https://datacenterfrontier.com/scalematrix-builds-niche-in-hpc-for-genomics-ai/