Dell HPC Scalable Storage

The Large Hadron Collider (LHC) research community needs inexpensive, high-performance accessible storage for its massive and ever-increasing research data pool. With 449 institutions across the globe working on the LHC project, many institutions need help implementing a mature storage platform specific to their function in the LHC research. Dell works with the LHC community to understand their storage requirements and provide the best possible practices and available storage hardware.

To study the bandwidth, capacity and cost of storage solutions used in Tier 2 and Tier 3 LHC environments, Dell teamed with Deutsches Elektronen-Synchrotron (DESY). DESY compared the dCachepool on two Dell scalable storage solutions ― the Dell™ High Performance Computing (HPC) Scalable Storage Building Block (HSSBB) and Dell R510 Storage Servers (RSS). DESY installed and submerged the Dell HSSBB and RSS appliances into a HammerCloud high-workload LHC test environment.

The DESY dCache test results showed that two HSSBBs pooled together delivered almost the same performance as that of five RSS. Analysis of results also produced the following conclusions:

• Many small storage nodes yield a stable solution for very large environments with large storage and core requirements.
• The R510-based RSS option scales very well in the DESY dCache environment.
• The cost against performance advantage goes to the RSS option.
• The high availability advantage goes to the HSSBB option.
• The Dell HSSBB solution implemented by DESY is a good starting place for Tier 2 and Tier 3 institutions that need to quickly set up a dCache option.
• The Dell LHC Team recommends exploring other solutions, such as a multicontroller server with several just a bunch of disks (JBOD) trays attached to it, to lower the cost and increase the performance of the data pool.