
ORNL recently partnered with Oracle to test some of the database technology company’s new firmware. Pictured here: Quinn Mitchell, high-performance computing storage system administrator at the OLCF, with the OLCF’s tape libraries. Image Credit: Rachel McDowell, ORNL
The US Department of Energy’s (DOE’s) Oak Ridge National Laboratory (ORNL) recently partnered with Oracle to test some of the database technology company’s new firmware. Under this collaboration ORNL is using Oracle’s tape drives in a collaborative laboratory environment inside the Oak Ridge Leadership Computing Facility’s (OLCF’s) High-Performance Storage System (HPSS).
Oracle frequently releases new firmware patches—specialized software embedded into a piece of hardware—that require testing prior to public release. The OLCF, a DOE Office of Science User Facility located at ORNL, has recently entered into a lending agreement with Oracle to receive five of Oracle’s enterprise-class tape drives for firmware testing and other kinds of beta testing. Oracle is providing its drives and firmware while ORNL is providing the computational resources, networking, and manpower for the testing environment.
Because Oracle comes out with firmware patches on their drives every 6 to 12 months, they needed a friendly, heavy-use site like ORNL to preliminarily test the firmware before presenting it to the public. “Working closely with Oracle in this loan agreement and testing future-release hardware will allow the OLCF to influence and contribute to the enterprise tape ecosystem in a very direct way,” said Quinn Mitchell, high-performance computing storage system administrator at the OLCF. “This collaboration places us in a position to provide Oracle with meaningful feedback on product performance, benchmarks, anticipated problems, and feature requests.”
The OLCF’s feedback is important because these updates will ultimately benefit users of Titan, its flagship supercomputer. The firmware improvements can help reduce data loss and increase overall data performance. Oracle recently added read-ahead accumulation, a firmware feature that rereads the tape multiple times if it finds an error and acquires data the drive can read in the process. This feature can prevent the data loss associated with retrieval errors—especially for older data.
These kinds of firmware tests, Mitchell said, can provide the OLCF with extra data integrity sooner. “This collaboration is beneficial for Oracle, but it’s beneficial for us, too,” Mitchell said. “Ultimately it can help greatly improve media degradation problems.”
Oracle is also working with the OLCF to build tape drives that have connectivity via remote direct memory access over converged Ethernet (RoCEv2), an interconnect that is simpler and less expensive than traditional Fibre Channel (FC), a complex data storage network technology. The integration of RoCEv2 into the OLCF’s HPSS would eliminate its current FC infrastructure.
“Substantially less money goes into Ethernet than FC, so RoCEv2 could save thousands in server builds for tape and storage infrastructure costs,” said Mitchell.
Oak Ridge National Laboratory is supported by the US Department of Energy’s Office of Science. The single largest supporter of basic research in the physical sciences in the United States, the Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit science.energy.gov.