[Bioclusters] Nomenclature (was Re: Call for information.)

Ivo Grosse bioclusters@bioinformatics.org
Thu, 18 Apr 2002 22:23:15 -0400


Hi Goran, hi all,

could you please clarify:

50-60 MBytes/s or 50-60 Mbits/s?

50-60 MBytes/s would give 400-500 Mbits/s, so that would mean that a 
Gigabit connection could increase the throughput by a factor of 4-5, in 
an ideal world, correct?

What is the situation for an IDE RAID array?

Is there any hope to get a 1-TB fileserver for less than $10^4 that 
gives I/O speeds > 1 Gbit/s?  Chris?  Others?

Best regards, Ivo

 
Goran Ceric <goran@genetics.wustl.edu> wrote on Thu, 18 Apr 2002:

> Not true. It's more like 50-60 MB/s. 
> 
> > 
> > Now someone mentioned to us that the physical speed of SCSI disks is of
> >  the order of 100 Mbits/s.  If that is true, then of course a Gigabit 
> > connection to the fileserver makes no sense, because the bottleneck 
> > will be the r/w speed of the disks, and not the network.