I just got a 24 port 1GB switch and my first attempt definitely didn't give the results I was hoping for. Perhaps there are some tuning options that might improve the options. First, I created an ntfsclone lzop compressed image of the Windows partition that is uploaded to one of the machines connected with the 1GB switch. That took about 11 minutes. I then made sure that all the machines had the nics running at 1G. Originally, two machines were only connecting at 100M, but replacing the connector on 1 and just recrimping the other got them all to connect at 1G. The cables are all Cat5e, but they are a few years old. I have a script that using udpcast to transfer this image file to all the other 19 machines in the lab connected to the machine. I was hoping to get a better speed when switching to the 100M to the 1G, but wasn't looking for a 10 times increase, since I knew that the hard disks only will do about 55MB/ second accordng to hdparm. But in my first test it took 31 minutes to transfer the file using udpcast?? Image file is about 12GB. Here is the end of the listing. bytes= 12 221 254K re-xmits=32506028 (378.1%) slice=0112 - 15 Timeout notAnswered=[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18] notReady=[0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18] nrAns=0 nrRead=0 nrPart=19 avg=252 bytes= 12 229 698K re-xmits=32529469 (378.2%) slice=0112 - 3 Transfer complete. Didn't notice any machine giving errors, since it seemed to always list all 19 machines. The re-xmits seems to be an issue, but perhaps some other setting with the Linux (Fedora 16) or udpcast options. The script is run from the machine machine, and it links to the other 19 machines and then transfers. Partition /dev/sda6 is is a ext3 partition that is separate on all machines to store images, that can then be used to reimage individual machines with no network traffic. if [ -e /sda6/d7-xp.lzo ] ; then echo "SDA6 already mounted" else sudo mount /dev/sda6 /sda6 fi cmd="hostname && mount /dev/sda6 /sda6 && cd /sda6 && ./udp-receiver --no-progress --file d7-xp.lzo && cd / && umount /sda6 " #cmd="udp-receiver --file d7-xp.lzo " plink -t 192.168.7.201 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.202 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.203 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.204 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.205 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.206 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.207 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.208 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.209 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.210 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.211 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.212 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.213 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.214 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.215 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.216 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.217 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.218 -l root -pw xxxxxx "$cmd" & plink -t 192.168.7.220 -l root -pw xxxxxx "$cmd" & sleep 20 cd /sda6 time udp-sender --full-duplex --file d7-xp.lzo --min-receivers 19 --min-wait 9 I'm hoping there are some more experience with using udpcast in 1G environments. Currently running another test with sending to just two machines, but seems be showing similar results. +----------------------------------------------------------+ Michael D. Setzer II - Computer Science Instructor Guam Community College Computer Center mailto:mikes@kuentos.guam.net mailto:msetzerii@gmail.com http://www.guam.net/home/mikes Guam - Where America's Day Begins G4L Disk Imaging Project maintainer http://sourceforge.net/projects/g4l/ +----------------------------------------------------------+ http://setiathome.berkeley.edu (Original) Number of Seti Units Returned: 19,471 Processing time: 32 years, 290 days, 12 hours, 58 minutes (Total Hours: 287,489) BOINC@HOME CREDITS SETI 11801304.490587 | EINSTEIN 7206065.819851 ROSETTA 4194793.509365 | ABC 10747635.646984