Posts

Showing posts from October, 2008

Links to NAS and SAMBA performance

In my prevoius blog I wrote about my tests to increase the read/write speed on my Linux home server, now I will collect some links here: A very good comparition of Windows and Linux file system for using as a NAS. Test result for a consumer NAS device. Test results for Linux file systems (ext2, ext3, Reisser, XFS). IOZone test suite. NAS performance measurement comparations. A test measurement of comparing Solaris ZFS, Ext2 and XFS in NAS environment. It is using NFS intead of SAMBA, but the speed results are very convincing with read speeds over 76 Mbyte/sec and write speeds arround 35 Mbyte/sec. I started to do some own test in the server itself, and copying a file from one disk to an other was done with the speed of 23,4 Mbyte/sec, which shows, that it is possible to write to an ext3 filesystem with speeds above 20 Mbyte/sec. I will do some more testing later. I did some testing on my Windows Desktop computer and got a read and write speed of ~90 Mbyte/sec, and thi...

NFS Client for Windows XP computer crashing with large files

I have recently installed a gigabit card to my home fileserver and was very dissapointed, that the speed of network writes and reads did not increased as I expected. I found some Samba improvement tips, but they increased the speed only by ~10 percent, while I would expect to have network speed on the range of 50 Mbytes/sec. To make an experiment I wanted to use NFS instead of Samba, from one unix test it looked much faster than Samba. There is an official Microsoft client for this, it is named Unix Services for Windows, the actual version is 3.5 and it is quite outdated. Still after installing it and configuring both the server and the Windows XP it worked adn when attempting to read files it was quite stable. But as soon as I tried to copy a 4,3 Gbyte ISO image, my XP crashed completly :-( After a lot of search finally I could identify that you should apply the hotfix: KB904838 and this will solve your problem. Back to the original problem, I have measured the following sp...