| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1176.1 | | PSW::WINALSKI | Meetings are our most important product | Wed Jul 26 1989 13:27 | 6 |
| The ethernet here at ZK is probably the busiest in the world, and the use of
DECwindows over the network has not caused appreciable problems here. The DR
article is pure viscious rumor-mongering.
--PSW
|
| 1176.2 | | VWSENG::KLEINSORGE | XUIS or DIE. | Wed Jul 26 1989 16:43 | 9 |
|
Now Paul, ZK probably also has some of the heaviest-duty talent keeping
the ethernet working -- and it didn't (doesn't) always work well. It
also uses a lot of strategic bridges which help to reduce the traffic.
AND maybe most important, it DOESN'T have 30 people doing heavy-duty
CAD/CAM (or other graphic intensive) applications with the client on a
6240. 30 people running DW-MAIL is pretty insignificant.
|
| 1176.3 | The grossly over-simplified answer is "no problem"! But... | MCNALY::MILLER | Can I have a haircut like Sinead's? | Wed Jul 26 1989 17:18 | 14 |
| re: Do we have actual measurement on traffic for decwindows networks?
Bruno,
Check out Dinesh Mirchandani (pool::dinesh) and Prabuddha (glivet::)
Biswas ' paper on "Characterizing and Modeling Ethernet Performance
of Remote DECwindows Applications in an LAVc". It goes into great depth
on what the impact of DECwindows applications like DECwrite will have
on the Ethernet.
Regards,
== ken miller ==
|
| 1176.4 | grazie! | ROM01::CIPOLLA | DEC's margin on an IBM sale is zero! | Thu Jul 27 1989 06:16 | 12 |
| RE: .-1
>Check out Dinesh Mirchandani (pool::dinesh) and Prabuddha (glivet::)
>Biswas ' paper on "Characterizing and Modeling Ethernet Performance
>of Remote DECwindows Applications in an LAVc". It goes into great depth
>on what the impact of DECwindows applications like DECwrite will have
>on the Ethernet.
>
thank you very much ken!
Bruno
|
| 1176.5 | Re-read the article | POOL::BUFORD | It's the Bill of Rights, not wrongs | Thu Jul 27 1989 11:07 | 8 |
| Go back and re-read the article and pay attention to what they did to
"alleviate" the load that DECWindows was supposed to be putting on the
network: local paging disks. Now, what does a local paging disk have
to do with X protocol traffic???
John B.
|
| 1176.6 | | PRNSYS::LOMICKAJ | Jeff Lomicka | Thu Jul 27 1989 14:20 | 6 |
| Nobody said that "X protocol" was loading the network. What they said
was that "DECWindows" was loading the network. That the offending
portion of DECWindows was the the virtual memory requirements, rather
than the protocol, is insightful, but it's still DECWindows that is
causing the increase in network loading.
|
| 1176.7 | Well, to get really sticky about it | VMSDEV::HALLYB | The Smart Money was on Goliath | Thu Jul 27 1989 18:00 | 8 |
| (Fake huffery here) Actually, it isn't DECwindows that's causing the
network loading, it's the memory requirements of DECwindows that's
causing the network loading.
Important if the systems have expandable memories...
John
|
| 1176.8 | The message is misleading and dangerous | POOL::BUFORD | It's the Bill of Rights, not wrongs | Fri Jul 28 1989 13:34 | 28 |
|
re .6:
> Nobody said that "X protocol" was loading the network. What they said
> was that "DECWindows" was loading the network.
Precisely. What they said is misleading and can cause someone to apply
the wrong "solution."
The message "DECWindows loads down the network" implies that I should
run my applications locally in order to minimize network traffic due to
X protocol. *WRONG* In fact, I have three alternatives:
Add more memory to the workstation
Use a local disk for paging
Run clients on a remote node that has sufficient memory to
minimize paging
That third option is the kicker: to reduce network traffic, I can run
my DECWindow apps *remotely*. That makes no sense if you believe what
that article what trying to say...
John B.
|
| 1176.9 | a pointer to the DECwindows/Ethernet paper | VMSDEV::DINESH | VMS Performance Tools | Fri Jul 28 1989 16:02 | 25 |
| >>> re: Do we have actual measurement on traffic for decwindows networks?
Bruno,
>>> Check out Dinesh Mirchandani (pool::dinesh) and Prabuddha (glivet::)
>>> Biswas ' paper on "Characterizing and Modeling Ethernet Performance
>>> of Remote DECwindows Applications in an LAVc". It goes into great depth
>>> on what the impact of DECwindows applications like DECwrite will have
>>> on the Ethernet.
____________________________________________________________________________
hi, this paper is available on the network at:
VMSKIT::SYS$PERFDOCS:DWLAVC.PS
it is 70 pages long, 5300 blocks of postscript.
i hope that you will find answers to some of your questions in the material
presented here.
regards-
dinesh mirchandani.
vms performance modeling.
|
| 1176.10 | Don't hide the candle under a basket!! | CSC32::B_WACKER | | Fri Jul 28 1989 17:21 | 3 |
| Will someone take the time to set DEC Professional straight? I think
they'd be open to a responsible rebuttal.
|
| 1176.11 | DEC Pro is usually correct to begin with. | DWOVAX::YOUNG | Another pushy field person. | Mon Jul 31 1989 02:49 | 5 |
| Re .10:
I agree, the DEC Professional would be open to a responsible rebuttal.
Unfortunately it was Digital Review who published this...
|
| 1176.12 | | XUI::VANNOY | Jake VanNoy | Mon Jul 31 1989 11:37 | 9 |
| I talked to DR about this on friday. They are very willing to set the record
straight, and feel that the "customer impressions" are more from lack of
knowledge than anything else.
DR is working on a special DECwindows issue for later in the year, we'll try
to get a performance/overhead related article in that issue.
jake
|
| 1176.13 | | XUI::VANNOY | Jake VanNoy | Mon Jul 31 1989 11:54 | 10 |
| Oops, I met with Digital News on friday, not Digital Review. D/News has
published some unfavorable things too (like being "far from the mainstream
of X").
We gave them about 3 hours of demos and Q&A on friday. They went away a lot
smarter about DECwindows than when they arrived. One of the writers kept saying
"I want this, I want this in my office".
jake
|