Test
After having analysed performances of NFSv4 for files creation, and compared it to ext3 performances, the next step is to compare NFSv4 to NFSv3 for files creation. Scripts used are the same for all tests. Script explicitly specifie to create and sync the file (open with O_CREATE and fsync(fd)).
Diagrams:
- The first diagram is the time to create n files on NFSv3 and NFSv4.
- The second diagram is the time to create n new files on NFSv3 and NFSv4
While we can see NFSv3 and NFSv4 have the same behaviour for file creation for a low number of files, NFSv3 is not affected by the limitation of 1620000 of NFSv4.
diagram:
- These curves represent a detail of the above curve. It represent the time to create 100 new files in a directory for NFSv3 and NFSv4.
- We can see the same behaviour of the dispersion of data for NFSv3 and NFSv4 when number of files in directory is between 160,300 files and 734,500 files.
- Since this zone repartition of the data dispersion does not affect local file system (see here) we can suppose this behaviour is caused by the core NFS subsystem (NFSv3-NFSv4 common parts).
Conclusion
- Limitation of file creation is caused by NFSv4 it self
- Some behaviours of NFS files creation are common to NFSv3 and NFSv4
Software configuration
- Server :
- Export options are :
/export *(rw,wdelay,insecure,no_root_squash,no_subtree_check,fsid=0)
- mount options, NFSv4 :
rsize=32768,wsize=32768,timeo=14,intr,noauto
- Linux 2.6.12-rc4-CITI_NFS4_ALL-1
hardware configuration
(Client and Server)
- 2 processors : Intel(R)
- Xeon(TM) CPU 2.80GHz, cache 512 KB
- Total memory: 2Gb
- Ethernet: 1Gb/s link
- Distribution : modified Fedora Core 2
