| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 2963.1 | TRANSMITTERS not receivers | MARVIN::RIGBY | No such thing as an alpha beta | Thu Jan 25 1996 05:55 | 21 |
| 2963.2 | Max Queue Length = 0 ? | MARVIN::GLANVILLE | | Thu Jan 25 1996 07:44 | 10 |
| 2963.3 | still have problems, questions | PRIME3::THOMAS | | Thu Jan 25 1996 15:53 | 29 |
| 2963.4 | why no boot with priority defined? | PRIME3::THOMAS | | Thu Jan 25 1996 16:57 | 8 |
| 2963.5 | Prioritisation not supported on FDDI - Filtering is | MARVIN::GLANVILLE | | Fri Jan 26 1996 07:23 | 26 |
| 2963.6 | | MARVIN::MCCLURE | Tony McClure, DECnis Engineering RE02 FD9 7830-3564 | Fri Jan 26 1996 08:52 | 34 |
| 2963.7 | When is a circuit too fast to use prioritization | SCASS1::ROCHELEAU | Rod Rocheleau - 972-527-2272 | Wed Feb 19 1997 20:29 | 25 |
| >> Each prioritastion filter added increases the number of tests
>> done on this frame, however the over-head of these is quite small..
>> But if you have a lot of filters these tests do mount up..
>>
>> If the result of the prioritiastion is filter is pattern dependant
>> then the length and the number of patterns tests will again
>> produce quite a lowering if maximum performance. Pattern
>> filters require many more thst and therefore present a heavier
>> over-head.
This is a little off the original subject but ... We are trying to determine if
there is a break point on a point-to-point circuit where priorization is a
costing more in performance on the hardware than we are gaining on the circuit.
The would only apply to high speed circuits ... We would assume that a T1/E1
would benefit from Telnet (port 23) and LAT (bridge type 60-04) being
prioritized as 2 and 1 with everything else left to fend for it's self. The
desired results is to promote telnet to near LAT performance on the wide area.
Any guide lines would be helpful ... i.e. how much overhead does it take
if a 3.0mb pipe is used running an average packet size of 200 bytes?
Thanks in advance
Rod
|