[UniMacTech] Aggregated Network Ports

[Raoul Callaghan]

Hi Peter,
I too have set up some Xserves here using aggregation.
I used SmallTree Nics to achieve this about 2 years ago.  (2 x 2 port,  
and 1 x 4 port)

First of all, let me say:
A single SATA 7.2K rpm  drive will starve even a single Gigabit  
connection so go stripe some drives within those MacPros if you  
haven't done so already.

Okay. What you are seeing is normal.
The bond interface will only utilise the additional copper when the  
server receives additional requests. a.k.a high load...

This means that:
You will not see an increase in speed with only a single machine  
accessing the server, but rather you'll notice the bond interface  
being utilised once additional server requests come in.

So, I guess you could say that you'll only notice the same transfer  
speeds when the server is under considerable load: and therefore  
maintaining an optimal transfer speed for you.

If you have spare Nics lying around, you can simply use these, (with a  
different IP though) as AFP, like Apache, is set to listen on all  
interfaces by default.

Once you start playing around with bond interfaces, you quickly  
appreciate RAID 50, as drive spindle speed soon becomes your new  
bottleneck. both on your clients as well as your servers.

Cheers,

Raoul Callaghan
I.T. Manager
Australian Mathematical Sciences Institute
111 Barry Street
The University of Melbourne
Victoria 3010 Australia
p: 03 8344 1783
f:  03 9349 4106
e: raoul at amsi.org.au
  Aoccdrnig to a rscheearch at an Elingsh uinervtisy, it deosn't  
mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt  
tihng is taht frist and lsat ltteer is at the rghit pclae. The rset  
can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is  
bcuseae we do not raed ervey lteter by itslef but the wrod as a wlohe.


On 01/04/2008, at 6:39 PM, Peter Thomas wrote:

> Dear All,
>
> I have just installed a new gigabit Cisco switch to upgrade network   
> speed in  my teaching labs and studios. The switch has some ports  
> configured to allow for aggregating or trunking ethernet cards on my  
> servers and some of my Mac Pro studio computers to provide faster  
> transfer of large files and faster image deployment. (Also maybe  
> play with Final Cut distributed rendering).
>
> I have aggregated, (bonded) the two network ports on a Mac Pro tower  
> and done some rudimentary speed tests on this and another identical  
> computer that is configured to only use one network port.
>
> My test basically was to copy the same 2 gig file across the network  
> from our server on both computers simultaneously and time them –  
> expecting the computer with the aggregated network ports to be twice  
> as fast – not so IDENTICAL
>
> I also copied the files back from the two machines back to the  
> Server – again expecting the computer with the aggregated network  
> ports to be twice as fast – not so IDENTICAL
>
> So how would I test whether I am achieving faster network  
> throughput, on the aggregated machine, is my test invalid or is  
> something wrong with switch or computer setup ?
>
> Thanks
>
>
> Peter Thomas
>
> Information Technologies Facilities Manager
> Conservatorium of Music
> University of Sydney
>
> _______________________________________________
> unimactech mailing list
> unimactech at auc.edu.au
> http://www.auc.edu.au/mailman/listinfo/unimactech

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://auc.uow.edu.au/pipermail/unimactech/attachments/20080402/20f0f85d/attachment.html