X-ISS Manages Dell Life Science HPC Solution with Complex Security Protocols
An organization engaged in cancer diagnostics needed to leverage HPC cluster technology to provide for quicker turnaround time to end users. At the same time, not only did they need to ensure the clusters were fully managed but also to abide by complex security and validation protocols to maintain patient accuracy and confidentiality in clinical trials. Mandated at the federal level, these protocols directly impacted how the organization’s HPC was deployed and used on a daily basis. X-ISS was contracted to manage the clusters and help ensure all confidentiality and regulatory guidelines and regulations were followed.
According to a key stakeholder, “X-ISS was instrumental in enabling and supporting the underlying infrastructure running a critical set of cancer diagnostic tests, thereby alleviating a significant resource burden from within the organization. The expertise and willingness to operate within the necessary regulatory environment significantly impacted the outcome in positive ways.”
Dell introduced an HPC cluster infrastructure specifically to serve the needs of Life Sciences organizations, and this was the first one sold. Among its many special features, the design included a Lustre file system for high-speed file writing and NFS for long-term file archiving. This accommodated the data processing demands common among Life Science organizations.
“A more typical HPC cluster solution is designed to handle a small number of jobs with extremely large files, as is often required for engineering and manufacturing organizations,” said X-ISS CEO and Founder Deepak Khosla. “Life Science organizations, on the other hand, usually run numerous jobs with a lot of small files written to storage, and that’s how Dell designed this cluster.”
The client in this case had to abide by federal HIPAA, PII, and Food and Drug Administration (FDA) CFR21 regulations which govern the security that must be provided to patients involved in clinical trials. Dell customized its off-the-shelf Life Science solution for the client by partitioning a single rack of nodes into four logical clusters. The clusters shared common back-end high-speed and long-term file storage.
The four clusters included one for development, one for validation, and two for production. The two production workflows had to be separated so that sensitive clinical trial data could be processed in a restricted environment where a limited number of cleared personnel had access in accordance with the confidentiality rules.
The client contracted X-ISS under the Dell Remote Cluster Management service to assist with deployment and then manage three of the four clusters, excluding the development cluster. Before starting, X-ISS personnel underwent the same HIPAA and regulatory training sessions that all of the client’s regular employees completed upon joining the company.
Among the most important deployment services provided by X-ISS was assisting with the installation of applications on the clusters. This had to be accompanied by exhaustive written validation documentation of the installation process, which had to comply with the various mandated regulations.
“Our team first installed the applications on the development cluster,” said Khosla. “Once that was completed, and all of the written processes verified, the X-ISS personnel repeated the installations on the validation and production clusters.”
The formal documentation was required for federal auditing purposes. If the FDA were ever to audit the deployment and set-up of the clusters, the client would be required to show the written documentation proving the applications had been installed according to approved guidelines.
Some of the installed applications included the following:
- BWA (Burrows-Wheeler Aligner)
- BCL2FASTQ
- GATK (Genome Analysis Toolkit) with Appistry CGA (Cancer Genome Analysis)
- R Statistical Package
Other important set-up projects the X-ISS team worked on also pertained to security. The client wanted to control authentication from the IT side. So X-ISS configured the clusters to authenticate against the customer’s enterprise Active Directory. In addition, X-ISS changed the default storage configuration to prevent patient information from being shared between the clusters.
In the final stage of deployment, X-ISS customized the scheduler configuration to allow limiting jobs by available memory resources. This put the client in greater control of its clusters, allowing applications and projects to use only specifically allotted memory. This ensured the pipeline would always be available to handle and complete the number of jobs expected in a given period of time.
With the configuration and deployment finished, X-ISS has continued to provide daily remote management services that have kept the client’s clusters running efficiently as vital clinical trials are conducted in support of cancer research.
Download this case study:LifeScience.CaseStudy11