T.R | Title | User | Personal Name | Date | Lines |
---|
673.1 | | GBI01::CARUSO | Angelo Caruso - SWAS Genoa, Italy | Sat May 16 1992 19:22 | 4 |
673.2 | I need them too... | FILTON::BIRCH | Please Mr Music, will you play? | Mon May 18 1992 10:10 | 9 |
673.3 | Try LEDS""::MARKETING:[PUBLIC] | SCAACT::HILDEBRAND | Help find the VUPsuckers! | Tue May 19 1992 10:01 | 0 |
673.4 | Thanks !!! | GBI01::CARUSO | Angelo Caruso - SWAS Genoa, Italy | Sun May 24 1992 16:59 | 1 |
673.5 | Which version is the latest? | APACHE::ROY | I don't drive fast, I fly low | Wed May 07 1997 12:38 | 10 |
|
re: .3
I looked in [public.docs], and found 6 versions of the
EK-HZFAM-UG.PS Does anyone know which of these is the most recent?
Also, does it cover anything about the HSZ50? I'd rather not waste
bandwidth and disk space trying to find out. Not to mention printing
them out.
thanks, Glenn
|
673.6 | ah, VMS!!!! | PCBUOA::WHITEC | Parrot_Trooper | Wed May 07 1997 16:07 | 6 |
|
Hey Glen.......dir/date might work?
Hows things?
chet
|
673.7 | Input, I need more input!!!! | APACHE::ROY | I don't drive fast, I fly low | Thu May 08 1997 08:33 | 33 |
|
Hi chet. Things good! In MKO heading for ZKO.
dir/date....... DOH!!!!!!! Been 3 years since I've even thought
about VMS, much less touched it! Let's see now, how do I get the
parameters on this file????? 8*)
Anyhoo, these versions are 1994. Three years ago. Does anyone
know where todays 'versions' of the same thing are? Such as HSZ50
users guide, and/or any other technical pieces relating to disk
subsystems. I do have quite a bit now from the SOC and
StorageWorks web sites. I know there's a ton of info out there,
I just can't find what I want.
wish list
- up to date SCSI overview. right up to Ultra.....
- overview of DEC approach to designing disk matrices (?)
- config guidelines
- performance vs configuration discussions. # controllers, etc....
In other words, I'm trying to figure out what to use to build the
system that I'm asked for, BUT, I want to know why I'm using the
parts. Why is one controller better than the other? How do I make
this thing 'rock' and get the performance that I can use to get an
edge in the market?
One of the problems we're seeing is that we seem to be I/O bound.
that's my story, and I'm sticking to it 8*)
Glen(n)
|
673.8 | speed differences..... | APACHE::ROY | I don't drive fast, I fly low | Thu May 08 1997 10:42 | 23 |
|
Update. I did access the whatsa.???? web site mentioned, and
there's a lot there.
In trying to sort out some of the physical differences, I came up
with the following. Is this true?
8 Bit 16 bit
ULTRA SCSI 20MB/sec 40MB/sec
FAST (Fast Scsi) (Fast Wide Scsi)
10MB/sec 20MB/sec
SCSI (Wide Scsi)
5MB/sec 10MB/sec
This is the way that I think I understand the differences. Am I
missing anything here?
thanks, Glenn
|
673.9 | | SMURF::KNIGHT | Fred Knight | Mon May 12 1997 12:15 | 32 |
| Those numbers are correct. HOWEVER, remember too that
those are the bus bandwidth numbers. So, consider if
you got 2 disks with exactly the same platters, heads,
and spindle motor. The only difference is one has the
"SCSI 5Mb/sec" transfer engine, and the other has the
"ULTRA wide 40Mb/sec" transfer engine.
Is the performance of these 2 disks any different? The
answer is NO. Their performance is exactly the same.
The difference is in the bus idle time. The bus that
has the "ULTRA wide 40Mb/sec" disk will have 8 times more
bus idle time (because the data still moves at the same
rate to/from the media).
Now consider mutiple units on the same bus. IF it took 2
of the SCSI 5Mb/sec units to saturate that bus (0 idle time),
then it would require 16 of the ULTRA 40Mb/sec units to
saturate their bus (these devices would simply use up that
extra bus idle time).
But, this is also sort of a contrived example just to show
a point since in reality, all the new ULTRA disks also have
faster motors, faster analog electronics (the heads), and
increased density; which all together means they move the
data faster (not only on the bus, but to/from the media as
well).
Also remember that those numbers don't account for any of
the scsi overhead. The "real" bandwidth numbers are slightly
lower than the published "maximums".
Fred Knight
|