This problem has been reported many times in the past, and every case that I know of has boiled down to either a switch set in half-duplex mode, or a bad Win32 ethernet card (i.e. hardware/firmware problems on the card itself). The Win32 chapter of the manual documents a few of these problems.
On Friday 08 December 2006 20:42, Brian Jones wrote: > Thanks for the replies. See comments below. > > > On 12/8/06 12:10 PM, "Robert Nelson" <[EMAIL PROTECTED]> wrote: > > > I get similar times between all the combinations I test, Windows -> Linux, > > Linux -> Windows, Linux -> Linux and Windows -> Windows. > > > > However it did take some tweaking. The first thing to check is to see if > > you are getting a lot of TCP/IP errors. > > This is not the case. I am not getting errors and this is the same result > on multiple systems. > > > This could indicate one of adapters > > isn't running full duplex or there is some other configuration problem. The > > other thing is the various TCP/IP tuning parameters such as Window Size, > > etc. Another thing that can make a huge difference, if you are running a > > gigabit network, is use large packets. > > These are running at 100 full, no gig yet. > > > > Then there are all the apples and oranges issues. Are the machines being > > tested all on the same network segment, do they have equivalent hardware > > (disk speed can make a huge difference), etc. > > No they are not all on the same network segment. However, the one system is > a blade in and IBM bladecenter. Adding another blade running CentOS 4.4 and > running a backup sustains the speed of the other linux boxes. The Windows > Server blade on that blade center runs at the slower speed. > > Either way on the Windows boxes doing file copies or in the case of SCP to > the linux box running Bacula, the speed is exceptionally faster then using > the Bacula client for windows. This has been tested in the middle of the > day and the middle of the night noting a slight improvement at night, but we > are talking 100 KB/s increase at best. > > AV is also not the issue as one of the boxes is not running it and the speed > is the same. > > Thanks for your help. > > > > > > > -----Original Message----- > > From: [EMAIL PROTECTED] > > [mailto:[EMAIL PROTECTED] On Behalf Of DAve > > Sent: Friday, December 08, 2006 9:20 AM > > To: bacula-users > > Subject: Re: [Bacula-users] Windows backup to network speed issues > > > > Brian Jones wrote: > >> I have a question on the speed of backing up Windows 2003/2000/XP boxes to > >> disk across the network. > >> > >> For 2 windows boxes, here is the rates I am getting though it is the same > >> for all the windows boxes: > >> Rate: 432.1 KB/s > >> Rate: 528.4 KB/s > >> > >> For the Linux boxes, here are 2 rates I am getting but again, it is the > > same > >> across all the Linux boxes: > >> > >> Rate: 9556.8 KB/s > >> Rate: 8616.3 KB/s > >> > >> I did a test on one of the 2003 server boxes backing up one file that was > >> about 13 gig in size. The speed was as shown above for Windows boxes. > >> Using winSCP to copy that same file to the same box running Bacula, I have > > a > >> rate of 4000 KB/s. > >> > >> Is the windows Bacula client slow? Are others seeing better speeds? I am > >> fairly new to Bacula so any help would be appreciated. > >> > > > > Anyone please correct me if I am wrong, no egos in play here ;^) > > > > I see speed differences as well. There are several things that explain > > most of my differences. The load on the server that is running the > > client can have a big effect. I have one very busy web server that backs > > up considerably slower than my FTP server. I noticed as well that full > > backups of a not in use file system go much faster than a differential > > of a heavily used filesystem. > > > > Not Bacula faults, just the situation. I have tried to move my backup > > schedules to run when there is the least number of processes to interfer > > with the Bacula client. > > > > Also, keep in mind, the transfer rate is not just the network speed > > attained, but the rate at which the data is transfered via Bacula which > > includes verification, compression, disk reading on the client box, and > > disk writing on the storage box. > > > > In my testing a straight data transfer such as ftp always outruns > > Bacula, so comparisons are not valid. > > > > DAve > > > > ------------------------------------------------------------------------- > Take Surveys. Earn Cash. Influence the Future of IT > Join SourceForge.net's Techsay panel and you'll get the chance to share your > opinions on IT & business topics through brief surveys - and earn cash > http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV > _______________________________________________ > Bacula-users mailing list > Bacula-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/bacula-users > ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
