Kitz Forum
Broadband Related => Broadband Hardware => Topic started by: Weaver on August 07, 2015, 04:20:17 PM
-
Comments requested please on the following proposal :
If you are testing new modems or routers, or comparison testing, follow these recommendations in order to avoid letting DLM muck up the results by overpowering the differences between the kit with changes that would have been applied by DLM anyway.
1. First wait 30 mins, then
2. Turn off your existing kit for at least 30 mins, then back on again,
3. Make any measurements on existing kit, then
4. Wait for 30 mins, then
5. Turn off existing kit and remove
6. Wait for 30 mins, then
7. Plug in new kit and switch on
8. Make any measurements on new kit, then
9. Wait for 30 mins, then
10. If appropriate, make second set of measurements
11. Turn off new kit and wait 30 mins
As a check on the reproducibility of the test, could finally plugin existing kit and measure again, or even repeat the entire procedure.
(a) Don't perform these tests near dusk.
(b) Don't perform these tests when the weather is changing.
(c) Document the characteristics of your line; check your line's health and if there is any chance of there being a fault, write it up.
(d) Do not perform tests while powered from a UPS (paranoid).
(e) Keep mains cables away from ethernet cables and RJ-11 cables.
(f) Always use the highest quality cables possible - best RJ-11 cables (and Cat 6 or higher ethernet ideally SSTP or SFTP, and document what RJ-11 cables you used)
-
What a pain!!
I've said it, so you don't have to.
Please tell me if this isn't good enough to keep DLM at bay.
-
If the community of Kitizens believe s that there's any merit in the proposal, thrn we could come to say, "tested under KETP" and such.
-
Sorry Weaver but I don't think I'd have enough spare time to do that lot :no: and my kit is all powered via a nice big chunky UPS :)
-
I don't think there's any need for step 1. Instead here I would add power down the unit from the mains before unplugging the modem from the telephone line.
Add comment re step 2 - This is to ensure one complete 15 min DLM accounting period without any uptime. This period is needed to ensure the DLM classifies it as an unforced retrain event.
As regards step 3 Im unsure if 30 mins is needed. I would think 15 mins would suffice - DLM will catch _any_ retrain events. With regards to the step 3 period, the DLM will be counting the retrain event rather than time.
15 mins would also ensure that in addition to DLM, the bRAS also has chance to catch the sync rate for IPprofiles, woosh recording etc.
A very quick resync sometimes escapes notice by the bRAS monitoring (RAP (http://www.kitz.co.uk/adsl/DLM_system.htm#RAP)). Although RAMBO does monitoring for both the DLM and RAP - its the RAP part that is related to the bRAS and IPprofile, wilst the DLM functions are performed at the DSLAM.
I'm also of the opinion that if you know your line has been stable then the odd one is perfectly fine. If Im doing a straight swap out or just want to reboot my router after a config change, then sometimes I cant be bothered to wait.
-
@ronski - I hear you. We do what we can.
One idea might be an idea to stretch things out more, so the tester does a bit on one day, then continues with step n+1 much later, rather than being tied up for a whole day.
And I'm on a UPS too.
The idea behind this is to banish the phantoms which I believe have possibly ruined some of my comparison tests because of DLM interfering.
-
@kitz - I agree with all your improvements. Any changes to make it less onerous very welcome of course.
If we have some measure of support and reach consensus could we publish a later improved revision on the site somewhere?
Would that make it not an RFC, but a KFC?
:o
-
. . . but a KFC?
Kentucky Fried Chicken :D
-
^ that made me laugh :D
-
^ that made me laugh :D
We aim to please.
Just couldn't help it.
Any yea or nay voters, please shout out.
-
Revised test procedure incorporating suggestions from Kitizens
1. Begin
2. Turn off your existing kit for at least 30 mins, then back on again,
3. Make any measurements on existing kit, then
4. Wait for 15 mins, then
5. Turn off existing kit and remove
6. Wait for 30 mins, then
7. Plug in new kit and switch on
8. Make any measurements on new kit, then
9. Wait for 15 mins, then
10. If appropriate, make second set of measurements
11. Turn off new kit and wait 30 mins
As a check on the reproducibility of the test, could finally plugin existing kit and measure again, or even repeat the entire procedure.
(a) Don't perform these tests near dusk.
(b) Don't perform these tests when the weather is changing.
(c) Document the characteristics of your line; check your line's health and if there is any chance of there being a fault, write it up.
(d) Do not perform tests while powered from a UPS (paranoid).
(e) Keep mains cables away from ethernet cables and RJ-11 cables.
(f) Always use the highest quality cables possible - best RJ-11 cables (and Cat 6 or higher ethernet ideally SSTP or SFTP, and document what RJ-11 cables you used)