samba file transfer is different from server to server
 

hi,

i have 2 samba server running in my netowork is, i had to back up around 150 gb of data to them, during the time i noticed a change that one was copying and receing at around 50Mbps, while another was hardly at 13 Mbps, why was the difference, below are the details of both

50Mbps 13Mbps
processor p4 p4
ram 524 Mb 1Gb
Hdd 160 Gb 2TB(samba) 160Gb(OS for OS)
CentOS 6.2 6.3

the server with more ram supposed to be faster but why it is in reverse.



Please help me find the casuse resolve it.

You're asking us why you're network speed is slower but haven't given us any info.

If I put a superfast system behind a 10Mb/s-100Mb/s switch and put an old POS system behind a 1Gb/s switch the transfers to my old POS system are going to go much faster. This is a problem with network throughput and has nothing do with system resources

Ditto on the previous post.
Also, do they have the same NIC? The ethenet hardware/driver also has a lot effect on performance.

hi,

they both are same systems same NICs, i also check them connecting directly to a 1Ghz switch and my laptop to the same switch, then i tried copying the files. Then also same speed. That's why i'm asking why does the difference between two samba servers. They are not at all in different network now. Same connectivity same NIC. Is there any other reasons for samba other then network because they are now directly connected to 1Ghz switch.

thanks.

So you're saying that EVERYTHING is the same for the two servers expect one has more RAM.

In which case I suggest you packet capture a portion of the file transfer for each server, then see if the slowness is due to latency or packet loss.
With all the plugins you get in Wireshark these days that shouldn't be hard to do.

hi,

i'm new to wireshark, if u don't mind can u plz tell me do it for me.

thank you very much.

and if you don't try to learn wireshark, you'll always be new to it.

capture the traffic, then look to see if there are any delays in the smb transaction, or packet retranamissions (caused by packet loss).
you have a good server and a bad server. shouldn't be hard to compare the two.
probably also a good thing to see the ifconfig stats to see if youare dropping packets too.
if latency at the server is evident, then perhaps look at the cpu load, and all software versions involved in the datapath.
looking at the logs during the transaction wouldn't hurt either.

what you're asking has too many variables. I think the above is a good start.

hi,

i'll try and get back

thank you.