[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Dual Processors ?



PureBytes Links

Trading Reference Links

For the fastest disk performance, consider a RAID array.  Data is written
to and read from the RAID array in parallel, one bit per drive (in the
most sophisticated RAID configuration), rather than one byte at a time.
There are also configurations with full redunancy, so that all information
remains even when one drive fails.

Rob Lake
rbl@xxxxxxxxxxx

> 
> At 06:20 PM 3/30/99 -0600, stuart wrote:
> >In regards to hardware, does anyone know how much the new ultra-wide,
> >ultra-fast scuzzi's improve performance?  I have a P-2 450 with a 10,000 rpm 9
> >gig Seagate Cheetah with Adaptec's fastest scuzzi, and it seems that TS 4.0
> >runs all my programs well.  I have Advanced Get RT running as well as TS 4.0
> >I have at least nine charts running in AGet and 5 windows in TS with several
> >charts in each window with dual monitors through Win 98 with no delay.  I use
> >BMI sat, and as a means of comparison with posts previously, I got 3722 ticks
> >on DSP9M today (3/30).
> >
> >Midnight updates through BMI take about 10 minutes to stop working  the hard
> >drive.  It seemed to take about as long with my former 200 mghz with similar
> >scuzzi.  I only run  190 symbols, but I store 5,000 days of each, including
> >all tick data.  Would I see the hard drive work less with dual processors,
> >dual HD scuzzi's, in RAID array, or is the advantage mostly in backup on the
> >second drive?
> >
> >SB
> 
> The Cheetah is the fastest drive currently available. The very fastest 
> drive interfaces are the optical ones. Then differential SCSI, which is 
> twice as fast as single-ended SCSI. Frankly, though all this raw power is 
> nice, and the differential price to get the latest and greatest is quite 
> small, most accesses are sequential. With sequential access, the existing 
> disk caches do a good job of insulating you from the failings of the hard 
> disk system.
> 
> I doubt that the time it takes to do your midnight is disk drive dependent, 
> so further improvements in that direction will probably not show benefits 
> wrt Omega thrashing the disk.
> 
> Allan
> 
> "Nothing that I can do will change the structure of the universe. But 
> maybe, by raising my voice I can help the greatest of all causes -- 
> goodwill among men and peace on earth." - Albert Einstein
> 
>