Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

I have to test PVFS i/O performance on a big file. First I am trying on a small csv file contains 1 000 000 rows. My computing is simple : it consists to get the number of differents pattern in the file. But my question is much about my test configurations. I have a NAS in RAID 6, and I am thinking to use it to store my data. Because my NAS have few disks, I am thinking that I will gain in I/O performance, instead using the local disk of my server. Am I wrong? I am not sure that PVFS will use differents disks from a NAS in the best way.

At this time, I use only one server for my computing. I would like to add one more server in PVFS configuration in the hope of I/O performance will be improve. Someone could confirm my idea?

PS : Sorry for my English.


share|improve this question

migrated from stackoverflow.com May 20 '11 at 13:48

This question came from our site for professional and enthusiast programmers.

1 Answer 1

For testing and verifying it works, I guess it's fine.

If you're talking performance, by the time your run of out performance headroom on a single server and parallel file systems such as PVFS and Lustre start to become attractive, you're talking six figure sums. Plus you want applications that can take advantage of MPI-IO, and a high performance interconnect network such as Infiniband.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.