Abstract
One of the most demanding tasks which Computing in High Energy Physics has to deal with is reliable and high throughput transfer of large data volumes. Maximization and optimization of the data throughput are therefore key issues which have to be addressed by detailed investigations of the involved infrastructures and services.In this note, we present some transfer performance tests carried out at the INFN-CNAF Tier-1 center, using SLC4 64-bit Grid File Transfer Protocol (GridFTP) servers and a disk storage system based on the General Parallel File System (GPFS) from IBM. We describe the testbed setup and report the measurements of throughput performances in function of some fundamental variables, such as number of parallel file and number of streams per transfer, concurrent read and write activity and size of data blocks transferred.During this activity, we have verified that a significant improvement in performances of the GridFTP server can be obtained using 64bit version of Operating System and GPFS from IBM.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
