I'd love to know how you people all get such fast data speeds!
With a gigabit connection from my desktop computer to the bubba the max I have seen is 4.7 MBytes/s
With my laptop with a "g" wireless the best I get is 1MByte/s
Since upgrading the bubba a couple of days ago I now get 100KBytes/Sec on the wireless connection.
I do have squeezebox server running. That is about the only thing I use on the bubba. Using the bubba as a fileshare is impossible as it is too slow.
Regards
George
New user's registration have been closed due to high spamming and low trafic on this forum. Please contact forum admins directly if you need an account. Thanks !
Network and disk throughput
Re: Network and disk throughput
Hello to all,
Since I upgraded to version 2 I get only 2 to 3 Mb per second, before I had 6 to 6,5 Mb per second.
I love the update but would like to see the speed (with windows through samba) the same.
Regards,
Puma
Since I upgraded to version 2 I get only 2 to 3 Mb per second, before I had 6 to 6,5 Mb per second.
I love the update but would like to see the speed (with windows through samba) the same.
Regards,
Puma
Linux is like a wigwam - no windows, no gates, apache inside!
Re: Network and disk throughput
Faster data transfer (10 - 11mbytes/s) is possible for Linux clients connecting over a wired network. The CPU utilisation is lower with NFS, when mounted with the -udp option.
http://forum.excito.net/viewtopic.php?f=9&t=1908
http://forum.excito.net/viewtopic.php?f=9&t=1908
Re: Network and disk throughput
Just checking if anyone has found something interesting that might boost up the NIC performance of the Bubba2 a bit more.
Re: Network and disk throughput
I solved this issue by buying a Mac mini server:p
Re: Network and disk throughput
I try to rsync some truecrypt volumes from an external esata to a external usb (for a test). The Performance is terrible, around 2 MB/s (I crosshecked this, it is true, not a measurement fault):
test_4gb.tc
4666163200 100% 1.64MB/s 0:45:19 (xfer#1, to-check=17/18)
test_70gb.tc
73400320000 100% 2.34MB/s 8:19:20 (xfer#2, to-check=13/18)
There is no network involved, just one port of the bubba to another port!
test_4gb.tc
4666163200 100% 1.64MB/s 0:45:19 (xfer#1, to-check=17/18)
test_70gb.tc
73400320000 100% 2.34MB/s 8:19:20 (xfer#2, to-check=13/18)
There is no network involved, just one port of the bubba to another port!
Re: Network and disk throughput
Hello,
Maybe rsync is causing that?
I copy data to USB drive and NAS with script files over night with crontab which gives me about 6.3 megabyte per second.
Try a simple bash script like this:
#!/bin/bash
# Crontab -e used for automatic start
cp -prv /home/storage/extern/[drivename]/ /home/storage/data/...../
echo ”ready with copy”
Good luck
Maybe rsync is causing that?
I copy data to USB drive and NAS with script files over night with crontab which gives me about 6.3 megabyte per second.
Try a simple bash script like this:
#!/bin/bash
# Crontab -e used for automatic start
cp -prv /home/storage/extern/[drivename]/ /home/storage/data/...../
echo ”ready with copy”
Good luck
Linux is like a wigwam - no windows, no gates, apache inside!
Re: Network and disk throughput
Yes, thank you, with cp it is about 5 times faster, which is good.
I am actually quite surprised, as I used rsync with the "--whole-file" switch, which should in theory disalble the whole delta computation and checksum thing and perform similar to cp.
The reason I did not use cp was that it did not work earlier with files that big, but as far as I remember I had to tune the ulimit or somthing like that.
Anyway, this proves that the reason for the very poor performance was not the bubba or the I/O, but the rsync itself (which is optimized torwards network transfer rather than local copy).
I am actually quite surprised, as I used rsync with the "--whole-file" switch, which should in theory disalble the whole delta computation and checksum thing and perform similar to cp.
The reason I did not use cp was that it did not work earlier with files that big, but as far as I remember I had to tune the ulimit or somthing like that.
Anyway, this proves that the reason for the very poor performance was not the bubba or the I/O, but the rsync itself (which is optimized torwards network transfer rather than local copy).