T.R | Title | User | Personal Name | Date | Lines |
---|
1473.1 | Yes | NPSS::MDLYONS | Michael D. Lyons - Young enough and dumb enough | Fri Oct 14 1994 17:49 | 13 |
| Whether or not a PMD is UTP or not is unrelated as to whether it is
SAS or DAS. Some product implementations may have restrictions, and
offer only certain configurations (I.E. some adapters may only have UTP
SAS interfaces), but there is no real connection between the type of
PMD and the type of FDDI port.
Take for example, the devices which we make which support the
modular PMDs, GIGAswitch/FDDI for example; you can mix any of the
available PMD types (multimode fiber, single mode fiber, TP-PMD) on any
port.
See 1458.1 If the configurations which are available are not what
customers want, let product management know about it...
|
1473.2 | | 56821::STEFANI | Have the # for the Mars Observer? | Sat Oct 15 1994 19:43 | 9 |
| >> Is UTP considered a SAS connection? IF so, do we ever expect to get
>> DAS UTP?
Michael (.1) is right. We sell SAS UTP adapters only, but that doesn't
mean we couldn't build a DAS UTP. In fact, we've been trying to make a
business case for DAS UTP PCI controllers. If you feel you have one,
please make your requests known to George Nielsen (DELNI::G_NIELSEN).
- Larry
|
1473.3 | | STRWRS::KOCH_P | It never hurts to ask... | Sun Oct 16 1994 21:24 | 9 |
|
I've just joined the NE region Network Product Group as a Sr. Systems
Engineer. I think this is important and I'll be trying to put together
a case for George from the field so we can get it built.
What would be the lead time from the "go" signal to actual delivery? Is
it just a matter or electronics or is there driver support required? Or
is the fact that the controller is SAS or DAS hidden from the driver
and handled at the hardware level?
|
1473.4 | | 56821::STEFANI | Have the # for the Mars Observer? | Mon Oct 17 1994 08:42 | 21 |
| >>What would be the lead time from the "go" signal to actual delivery? Is
I don't know all of the issues, but it would definitely require a
relayout of the existing design. Also, FCC Class B testing will be one
of the bigger hurdles to overcome. For a better understanding of the
hardware issues involved, contact Ralph Pepe (SCHOOL::PEPE) who is the
responsible hardware engineer. For scheduling and the like, check with
George.
>>it just a matter or electronics or is there driver support required? Or
>>is the fact that the controller is SAS or DAS hidden from the driver
>>and handled at the hardware level?
It's more a matter of electronics. The drivers don't care whether the
card is SAS or DAS, UTP or MMF. There might be some firmware-related
issues, but I don't believe it would be far removed from the existing
support we have for DAS MMF.
Good luck.
- Larry
|
1473.5 | why DAS? | 56821::DEFCTB::melaragni | | Tue Oct 18 1994 08:33 | 6 |
| Re .0:
Why do you even *want* a DAS UTP? I can't understand that. What do you
plan to do with it that a SAS can't do?
bill
|
1473.6 | | STRWRS::KOCH_P | It never hurts to ask... | Tue Oct 18 1994 09:32 | 13 |
| In a computer room configuration with more than 2 systems where turning
system on/off is not a problem, it reduces a single point of failure.
I realize that a DAS UTP is a single point of failure (get 2 SAS and
dual home them for redundancy to avoid single point of failure), but it
also provides for reduced costs to the customer when multiple systems
are involved. If we have a DAS UTP and a single system fails, the ring
seals itself and life goes on. However, if we have multiple systems
with SAS UTP and a single DECconcentrator 900, the DC900 can represent
a single point of failure for the computer room network.
I know it may sound like convoluted reasoning, but a DAS UTP gives us
additional flexibility and full range of options instead of having to
make excuses for not having it when constructing solutions.
|
1473.7 | | 56821::STEFANI | Have the # for the Mars Observer? | Tue Oct 18 1994 13:39 | 41 |
| >>I realize that a DAS UTP is a single point of failure (get 2 SAS and
>>dual home them for redundancy to avoid single point of failure), but it
Woah...there is no such thing as "2 SAS and dual-home". Dual-homing
(the connection of DAS A and B ports to two concentrator M ports) is
an exclusive option of DAS nodes. If your system, operating system,
protocol stacks, etc. provide the option of installing two SAS adapters
and allowing one to be a failover link, that's a higher level feature
that is not provided for or concerned with in the FDDI standard.
>>also provides for reduced costs to the customer when multiple systems
>>are involved. If we have a DAS UTP and a single system fails, the ring
>>seals itself and life goes on. However, if we have multiple systems
>>with SAS UTP and a single DECconcentrator 900, the DC900 can represent
>>a single point of failure for the computer room network.
More or less, DAS UTP adapters, like DAS MMF, or DAS anything gives you:
1. The ability to function as a SAS by connecting only one of the
two ports.
2. The ability to dual-home when connecting both A and B ports to
two M ports of the same or (preferably) different concentrator
M ports.
3. The ability to connect directly to the dual-ring by connecting
the A and B ports to the B and A ports (respectively) of other
dual-attach nodes.
DAS UTP just let's you do it cheaper. Of course, the reason Bill is
asking is that most folks that run FDDI backbones use fiber and then
opt to running UTP wiring to the desktop. Since most desktop
installations are single-attach, why do you need dual-attach UTP? My
answer would be for the customers that DO want to dual-attach at the
desktop, or dual-home, or run a UTP-based backbone, but don't want to
spend more for fiber-based cabling, and are willing/able to stay within
the distance limitations of copper cabling.
Of course, how much $ there is to be made in selling DAS UTP adapters
is still in question, which is why we need legitimate business cases to
justify the need.
- Larry
|
1473.8 | | STRWRS::KOCH_P | It never hurts to ask... | Tue Oct 18 1994 16:45 | 9 |
| re: .7
Larry,
I agree with you absolutely. I am trying to understand what the
business could be for this.
Thanks,
Ted
|
1473.9 | | STRWRS::KOCH_P | It never hurts to ask... | Tue Oct 18 1994 16:46 | 4 |
|
If we had a white paper discussing a DAS UTP which could be given
to customers so we can understand the market, this would be a great
help. We could send it over the Internet, get responses, etc.
|
1473.10 | | 56821::STEFANI | Have the # for the Mars Observer? | Wed Oct 19 1994 09:31 | 8 |
| >> If we had a white paper discussing a DAS UTP which could be given
>> to customers so we can understand the market, this would be a great
>> help. We could send it over the Internet, get responses, etc.
We're pretty much swamped here trying to get the PCI board out, but
feel free to write something up. :-)
- Larry
|
1473.11 | UTP DAS is just cheaper DAS, maybe | 40222::PETTENGILL | mulp | Sat Oct 29 1994 01:00 | 31 |
| Everything that applies to DAS as we know it today would apply to UTP DAS
other than the fact that fiber is involved. However, if people are concerned
about cost which is why they want to go UTP, are they going to be willing
to pay a good premium to pay for the lower volume and increased complexity
of DAS?
UTP DAS will be cheaper than fiber DAS if
1. the fiber DAS option uses MOD PMDs.
or 2. the UTP DAS volumes are as high or higher than fiber DAS
Given that those who are willing to pay for DAS are also concerned about
other issues related to reliability, will the fact that UTP conduct the
power for lightning strikes and other accidents of nature and man, where
will they be using UTP DAS connections. Will they want UTP DAS between
systems within a room, but fiber DAS to the systems in a different room
or power distribution? If so, then this means three new variants:
fiber A, UTP B; UTP A, fiber B; UTP A, UTP B, or it means that that the
option must use MOD PMDs. The cost of options with MOD PMDs will generally
be higher than same options that don't use MOD PMDs, so it might mean
that a UTP DAS option might cost as much as a fiber DAS option that doesn't
use MOD PMDs.
So, if option price is the reason for UTP DAS, then that might not be realized.
On the other hand, if use of existing wiring is the justification, then
maybe this makes sense. But this would imply that when the wiring was put
in years ago before UTP had reached its current focus, extra UTP was installed
for some purpose other than for use in dual ring FDDI. If you are going to
have to install new wire, then installing fiber is little different in cost
and that will be much more flexible in the long run. 655 seems very unlikely
on UTP (but then 100 seemed unlikely on UTP a few years ago).
|