[BBLISA] Fwd: Moving 100 GB and 1.3 million files

Bruce Davis ayden at mixolydian.org
Thu Jul 22 16:54:34 EDT 2010


I agree with this question.  How are you moving the files between
systems?  NFS, SMB, FTP, SFTP, SCP, RSH, RSYNC, etc?  Some protocols
are better than others.


Bruce Davis

On Thu, Jul 22, 2010 at 14:52, Theo Van Dinter <felicity at kluge.net> wrote:
> How are you moving these files between systems?  Is there a process
> you can strace/etc to see what calls are being made?  Are files being
> overwritten/updated on the remote side or are these new files?  Can
> you narrow down the issue to the sender or receiver?
>
> Some things I'd be concerned about: does the process need to compare
> file data/do diffs?  are the writes syncronous/are there fsync() calls
> when writing the files out?  are there any obvious bottlenecks --
> multiple people have noted network is not it, but is there something
> else on the system that's limited?
>
>
> On Thu, Jul 22, 2010 at 2:43 PM, Ian Stokes-Rees
> <ijstokes at crystal.harvard.edu> wrote:
>>
>>
>> On 7/22/10 2:34 PM, Rudie, Tony wrote:
>>> As a couple of people have said, it's not the gear that's the
>>> problem, it's all those little files.  But following in your
>>> footsteps, doing the same calculation based on 1.3 million files, we
>>> get:
>>>
>>> 50% done = 650K files, in 20 * 3600 seconds = 9 files per second.
>>> That seems low as well.  I just unpacked a tar file with 1000 files
>>> in it in 3 seconds.
>>
>> I should have included those numbers in my first email.  Yes, I've
>> looked at those, and they seem stupidly bad, like the system is running
>> orders of magnitude slower than it should.
>>
>> What keeps frustrating me is that no one can describe to me or point me
>> at a web page that describes the tools and techniques required to figure
>> out what is going on.
>>
>> Ian
>>
>> _______________________________________________
>> bblisa mailing list
>> bblisa at bblisa.org
>> http://www.bblisa.org/mailman/listinfo/bblisa
>>
>
> _______________________________________________
> bblisa mailing list
> bblisa at bblisa.org
> http://www.bblisa.org/mailman/listinfo/bblisa
>
>



More information about the bblisa mailing list