I have tried NASLite-NFS and repeated the very low-level measure of packet and data throughput to the drive.
METHOD
The computer is still an old Gateway Pentium 133 Mhz, 70 Mb RAM and 1.2 QuantumFireball hard drive. The BIOS is from IBM and is very minimalistic compared to modern BIOS's. The network card in this machine is a $10 Linksys (T10/100) PCI. This machine is connected to my home network (not wireless) to a Netopia ADSL router.
My workstation is a home built machine with a Gigabyte motherboard (nVidia n-Force2 Ultra chipset), AMD XP 2100+ CPU, 512 MB RAM and a 160 Gb 7200 rpm Western Digital harddrive. The operating system on this workstation is Xandros OS2 Deluxe.
I measured packet and data throughput (transmission) using the KDE Performance Monitor and a graphical display.
RESULTS
Here is what I got when I copied and pasted from my workstation to the NASLite-NFS server....
File Size (Mb)----Packets------------Data
1------------------67-73--------------100
3.6----------------67-73--------------100-110
5------------------67-73--------------80-100
10-----------------67-73--------------95-110
The packet throughput is about the same as when I used NASLite-SMB (see first post above). However, the data throughput is about 10% higher than it was with NASLite-SMB. Quantitatively, 10% is not a huge increasing, afterall, it is only 10%. However, I do believe that the data throughput using NASLite-NFS is greater than NASLite-SMB. This is because the data throughput never reached the 100 mark with NASLite-SMB, while data throughput consistently reached the 100 mark with NASLite-NFS.
Note: I currently do not know the units in these graphs (kb/s?).
Interestingly, the packet and data throughput does not change when I repeat the copy-n-paste across my home network from my workstation to another similarly spec'd Gigabyte machine (i.e., 2200+ AMD XP CPU, 512 Mb RAM). This finding appears to indicate that the performance obtained is not limited by the low CPU and RAM of the NASLite file server. It would be interesting to see whether performance would be increased with different hard drive configurations (7200 rpm versus 15000 rpm; single IDE disk versus IDE RAID; single IDE disk versus SCSI RAID).
Because the numbers above were obtained from a line graph, they should be considered best estimates. This is my first time using the KDE Performance Monitor and there may be a way to get more precise numbers.
CONCLUSION
After brifely using NASLite-SMB and NASLite-NFS, I prefer the NASLite-NFS because of the higher data throughput, and because browsing the file server with the Xandros File Manager (XFM) seems to be more responsive (i.e., files appear quicker). The only negative I have experienced with NASLite-NFS is that when I execute the copy-n-paste command in XFM a small window appears, that indicates the progress of the task, that gets stuck (the progress bar never reaches completion) and just abruptly disappears when the task is complete. This doesn't not happen in NASLite-SMB (i.e., the progress bar moves from left to right as data is written to the file server). However, this is minor and I don't mind it at all.
Note: I do not believe that the small performance difference between NASLite-SMB and NASLite-NFS has anything to do with the NASLite product and should not disuade others from using NASLite-SMB. The results shown here are probably not applicable to anyone else's system. This information is very preliminary and the methodology in testing and measuring can be improved for a more accurate comparison of performance.
|