Hi,
I have been doing some performance testing for an application. The test involves loading about 500K records into a database from a text file . The read to write ratio is 1:4 (using iostat). The read is being done from a partition on an IDE disk whereas the write is happeing on a ultra wide SCSI disk. I have noticed that for a this test compared to ext2 or ext3, when using jfs swapping occurs continously (as reported by vmstat column si, so and the noise from the disk :-) ). In the test the only difference is the filesystem. When using ext2 or 3 swapping rarely occurs. I must admit that this one single test cannot be used as a basis for compaing the two different filesystems. Plus probably there is not enough emperical information in this email. However this loading process is very important for our application and with JFS it is slower as compared to ext3 in ordered mode. A benchmark paper (using iozone) seems to suggest JFS out performing ext3 in sequential writes. Any clues on what could be going wrong on my test system will be much appreciated? BTW the version of jfs is 1.0.14 and kernel version is 2.4.18. Thx -Wajih __________________________________________________ Do You Yahoo!? Yahoo! Health - Feel better, live better http://health.yahoo.com _______________________________________________ Jfs-discussion mailing list [EMAIL PROTECTED] http://www-124.ibm.com/developerworks/oss/mailman/listinfo/jfs-discussion
