Business needTo help fight cancer and other diseases,

TGen needs extremely scalable, reliable

and available high-performance

computing (HPC) nodes to develop

personalized medical treatments.

Solution

TGen tuned its system for Genomics

I/O demands by scaling its existing Dell

HPC cluster to include more servers,

storage and networking bandwidth

so that researchers can get the IT

resources they need faster without

having to depend on shared systems.

Benefits• Researchers can create more-

targeted treatments at least one

week faster

• Improves outcomes for more

patients, including children battling

neuroblastoma

• Supports 100 percent data growth

and increases storage density

• Boosts efficiency so staff can

dedicate more time and money on

treatments

• Improves TGen’s agility so it can

scale critical aspects of its HPC

cluster in just a few days

• Increased utilization for Genomics

applications

Solutions at a glance• High-Performance Computing

• Infrastructure Consulting

• Storage

• Data Center Networking

Improving patient outcomes with personalized medicineThe Translational Genomics Research Institute helps develop custom treatments one week faster by scaling its Dell HPC cluster to support 1 million CPU hours per month

Customer profile

Company Translational

Genomics Research

Institute (TGen)

Industry Biotechnology and

Science

Country United States

Website www.tgen.org

“Today, we help save more lives because researchers spend less time waiting for HPC resources. And it’s also easy for us to scale and customize our Dell Genomic Data Analysis Platform to support our unique requirements.” James Lowey, Vice President of Technology, Translational Genomics Research Institute

2

“I’m not aware of any other solution on the market that’s like the Dell Genomic Data Analysis Platform. It’s optimized for genomic workflows out of the box, and within a few days, you can install, configure and launch it into production.” James Lowey, Vice President of Technology, Translational Genomics Research Institute

The unsung heroes of countless families, TGen’s bioinformaticians, geneticists and computer scientists strive to help more people survive disease — including neuroblastoma, a rare and often deadly cancer that typically afflicts children. They do so by developing a custom treatment formulated to specifically target the patient’s diseased cells and minimize damage to healthy tissue. Each patient’s unique chemotherapy formula is based on analysis that compares a patient’s genome and other biological information with data about the disease, other patients and their treatments.

With data from a single genome measuring up to 4 TB, sequencing a patient’s genome took about two weeks and performing the additional analysis to create a custom treatment took at least seven days. Positive patient outcomes often correlate directly with the amount of data that researchers can analyze and the speed of analysis. That’s why TGen strives to maintain an HPC cluster that delivers the server performance, storage capacity, network bandwidth and availability that’s needed by all users, at all times. James Lowey, vice president of technology at TGen, says, “Our staff includes some of the top people working on cutting-edge genetic and bioinformatics research. We have an obligation to ensure that our employees always have access to the best possible IT equipment and infrastructure.”

Scaling an HPC cluster in just a few days Two years ago, TGen deployed an HPC cluster based on the Dell Genomic Data Analysis Platform. Increasingly complex workloads, weekly additions of new

software and an ever-growing volume of patients were pushing the cluster to its limits. Not only was it sometimes difficult to get available nodes, but also researchers sometimes had to wait days to move data from storage to nodes due to network bottlenecks. To really accelerate analysis and create personalized treatments faster, researchers needed a more scalable cluster that they could easily optimize, along with their applications, to increase parallelization and overall performance.

IT personnel extensively vetted available technology options from large and small companies, and they found the best choice was to scale its cluster with

When it comes to treating cancer and other life-threatening diseases,

speed and precision are key to a patient’s survival. To develop more-

targeted and more-effective treatments faster, especially for children,

the Translational Genomics Research Institute (TGen) needs high-

performance computing (HPC) systems to quickly run extremely

complex algorithms that analyze terabytes of genetic and molecular

data from the patient and from research databases.

Products & Services

Services

Dell Infrastructure Consulting Services

Hardware

Dell Networking S4810 switches

Dell PowerEdge M1000e chassis

Dell PowerEdge R620 and M420 servers with Intel® Xeon® processors

Dell PowerVault MD3460 arrays

Dell Remote Access Controller (iDRAC) with Lifecycle Controller

3

Dell servers, storage and networking equipment. “I’m not aware of any other solution on the market that’s like the Dell Genomic Data Analysis Platform,” Lowey says. “It’s optimized for genomic workflows out of the box, and within a few days, you can install, configure and launch it into production. Other HPC systems generally take weeks or months to get everything up and running.”

Keeping the focus on research rather than IT by tapping Dell’s HPC expertiseTo design and deploy its new technologies, IT personnel collaborated with Dell Infrastructure Consulting Services. “It’s not easy to find vendors that understand technology and our use cases,” says Lowey. “Dell Infrastructure Consulting Services is unique because its staff really understand HPC as well as our unique requirements. For us, Dell isn’t just a vendor. It’s a true partner.” The feeling is mutual. TGen is a Dell Children’s Cancer Care Strategic Partner and recipient of a Dell Strategic Giving Grant.

Commenting on the overall deployment of the new solution, Lowey says, “By using Dell Infrastructure Services to come in and manage all the grunt work, including testing, we freed up cycles for our IT staff. Once the Dell consultants were done, all our people had to do was manage the final configuration so our researchers could start using the cluster to crunch genomic data as soon as possible.”

Researchers can create a custom treatment at least one week fasterBy increasing the size and performance of its cluster, TGen researchers have the compute resources they need. And because they are in-house, researchers can fine-tune computing environments along with applications to optimize overall performance. As a result, researchers can now develop custom treatments at least one week faster. Not only have they reduced the time needed for genetic sequencing from 14 days to 10 days, but also they have accelerated

the analytical processes (that facilitate custom treatments) from 7 days to 4 hours.

Lowey explains, “Last month, for the first time ever, our researchers logged 1 million CPU hours on our Dell HPC cluster — significantly increasing the amount and speed of research. There’s no way we would have been able to handle that volume of computation before, even if we used a university’s HPC cluster along with ours.”

Helps improve the quality of life for more childrenEvery hour, a child is diagnosed with cancer. TGen can help more children survive, and improve the outcomes for people battling other diseases. “Having this dedicated HPC resource enables us to really dig down into the intricacies and the complexities of the human genome and have a very positive impact on pediatric oncology as well as other projects,” says Lowey. “So today, we help save more lives because researchers spend less time waiting for HPC resources. And it’s also easy for us to scale and customize our Dell Genomic Data Analysis Platform to support our unique requirements.”

Saves time and money with faster performance and highly available systems To increase compute capacity, TGen added 32 Dell PowerEdge M420 server blades with Intel® Xeon® processors to an existing Dell PowerEdge M1000e chassis. “We get the best price/performance with Dell systems,” says Lowey. “For example, we get 512 cores in 10U with Dell PowerEdge M420 servers. To get that kind of industry-leading power in that small of a footprint is huge.” IT personnel minimize downtime by remotely managing servers via the integrated Dell Remote Access Controller (iDRAC) with Lifecycle Controller. “When you have hundreds of computational nodes running

4

unique and wonderful algorithms, nodes occasionally crash or run out of memory,” says Lowey. “When that happens, it’s pretty inconvenient to hop in your car, drive 30 minutes to the data center, and power the node off and on. With iDRAQ in Dell PowerEdge servers, we can manage servers, troubleshoot issues and hard-reset nodes from virtually anywhere, including my living room. So we save quite a bit of time and we’re able to maintain 99 percent availability, systemwide.”

Accelerates performance with two networks: Ethernet and FibreTo minimize network bottlenecks and reduce delays caused by moving data between systems or slow communication between cluster components, TGen segregated traffic between two separate networks. An Ethernet network now facilitates cluster component communication, and for that, TGen uses Dell Networking S4810 switches at the top of the rack. “Networking components are really critical to high availability and fast throughput,” says Lowey. “That’s why we leverage the 48-port Dell Networking S4810 switches extensively for our 10 GbE connections. A lot of people are using them in the HPC community — and like us, really like them.”

An InfiniBand® fibre network, which is based on four Mellanox® SX6036 FDR switches, supports all other traffic including connections between TGen offices and its data center. Commenting on why TGen chose Mellanox technologies, Lowey says, “We use the

Mellanox InfiniBand switches for the same reason we use Dell Networking switches: they just work, so we don’t even have to think about them.”

Supports 100 percent annual data growth and saves space with 1 PB in one rack To keep pace with a 100 percent year-to-year data growth, TGen scaled its existing Terascala storage so it can hold 1 PB rather than 215 TB. “We added 1 petabyte of Lustre file system capacity in just one rack of space by deploying six Dell PowerVault MD3460 arrays,” says Lowey. “Not only is the solution very dense, which is great because data center space is always at a premium, but also it’s faster by 12 gigabits per second over our previous storage solution.”

Streamlines administrative tasksTo expedite cluster management and make it easy to provision and manage microclusters for specific jobs, IT personnel implemented Bright Cluster Manager® from third-party Bright Computing. “In the past, we used different open-source tools to manage clusters,” says Lowey. “We really simplified reconfiguration and improved flexibility with Bright Cluster Manager. It’s easy to use, and it integrates seamlessly with our Dell HPC cluster, which is crucial in saving time and keeping us all as productive as possible.”

Dell, the Dell logo Dell PowerEdge and Dell PowerVault are trademarks of Dell Inc. Bright Computing and Bright Cluster Manager are trademarks of Bright Computing, Inc. InfiniBand is a registered trademark and service mark of the InfiniBand Trade Association. Intel and Xeon are registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. Lustre is a registered trademark of Xyratex and/or its affiliates, Mellanox is a registered trademark of Mellanox Technologies, Ltd. Terascala is a trademark of Terascala, Inc. Dell disclaims any proprietary interest in the marks and names of others. Availability and terms of Dell Software, Solutions and Services vary by region. This case study is for informational purposes only. Dell makes no warranties – express or implied—in this case study. Reference Number: 10013443 © November 2014, Dell Inc. All Rights Reserved

View all Dell case studies at Dell.com/CustomerStories

“Dell Infrastructure Consulting Services is unique because its staff really understand HPC as well as our unique requirements. For us, Dell isn’t just a vendor. It’s a true partner.” James Lowey, Vice President of Technology, Translational Genomics Research Institute

For more information go to www.DellHPCSolutions.com