50M Think Broadband Ping Monitor Results (POST YOURS) - Page 271

craigj2k12 · 04-10-2013, 23:09

Quote:

Originally Posted by Sephiroth

Ha ha Craigie! An updated TBB would be instructive.

Thats my live graph.....

Sephiroth · 04-10-2013, 23:32

In that case what remediation have you tried? Like when was the SH1 last rebooted? I'm trying to see if this is a variant of what TH reported some months back and what Ravenstart studied on the VM forum.

craigj2k12 · 04-10-2013, 23:42

Quote:

Originally Posted by Sephiroth

In that case what remediation have you tried? Like when was the SH1 last rebooted? I'm trying to see if this is a variant of what TH reported some months back and what Ravenstart studied on the VM forum.

as you can see i rebooted about 9 this morning, it seems to be a firmware issue since they changed the upstream packet buffering, where the superhub cannot cope under high load (CMTS side)

Sephiroth · 04-10-2013, 23:59

Quote:

Originally Posted by craigj2k12

as you can see i rebooted about 9 this morning, it seems to be a firmware issue since they changed the upstream packet buffering, where the superhub cannot cope under high load (CMTS side)

You know more than I do Craig. Where did they change the packet buffering? From what to what? For the period you were not there, what high load was there? Were you doing anything between 14:00 & 16:00 as shown on the current live graph?

In the various studies performed on the minimum latency perturbatuions (R37 it was), buffering and delay were strong candidates. But the phenomenon was always cleared for a coule of days by rebooting.

craigj2k12 · 05-10-2013, 01:08

Just rebooted about an hour ago and now have 2 upstreams, seems to have done the trick

---------- Post added at 01:08 ---------- Previous post was at 01:04 ----------

Quote:

Originally Posted by Sephiroth

You know more than I do Craig. Where did they change the packet buffering? From what to what?

Dont know but if you remember the jitter decrease came with R36 after I raised buffer bloat to Mark Wilkin on the VM forums

qasdfdsaq · 05-10-2013, 02:47

They fixed bufferbloat?

Chrysalis · 05-10-2013, 09:29

Quote:

Originally Posted by qasdfdsaq

They fixed bufferbloat?

they improved it, I still had some on my shub1 when my service was cancelled but it was noticebly improved from older firmware.

craigj2k12 · 05-10-2013, 20:46

Quote:

Originally Posted by Chrysalis

they improved it, I still had some on my shub1 when my service was cancelled but it was noticebly improved from older firmware.

Exactly

Kushan · 05-10-2013, 21:47

So uhh, you guys want to see something cool?

It's not a VM connection, thankfully, but it does make me glad I'm not on ADSL.

That said, we did have some connection troubles around midday today:

A reboot didn't make any difference but it cleared in about 20mins.

MattGarner · 06-10-2013, 23:17

So yeah...

buckleb · 07-10-2013, 08:23

The good service I've been having lately has come to an abrupt end!

Thank goodness for 3G

Sephiroth · 07-10-2013, 10:03

Your stats would help so we can see whaty the upstream is doing.

buckleb · 07-10-2013, 13:04

Quote:

Originally Posted by Sephiroth

Your stats would help so we can see whaty the upstream is doing.

My downstream is flicking between 1 and 7 channels and 2 to 1 upstream... Looks like something is broken before my sh2. VM say it should be fixed by 7pm, which is about the time I get in, so let's hope!

Sephiroth · 07-10-2013, 13:36

Quote:

Originally Posted by Sephiroth

.

Hi Craig

Yep - that graph is displaying exactly the phenomenon that thenry described at length in his long running CRAWLEY thread (sorry - you'll have to search for that yourself if you're interested). There are two keywords: "Crawley" and "Ramping".

The same keywords, but certainly "ramping" on the VM forum should bring you to Ravenstar68's thread where this phenomenon is basically debugged.

If I recall correctly (it's some months ago this year), wirel;ess lay at the centre of things but I can't be sure.

Keep us posted.

craigj2k12 · 07-10-2013, 15:15

Yes, I even contributed to the thread, my opinion was (and still is) the buffer on the router filling up, its the only explanation for the continuous ramping. This is doubly confirmed by the fact that the routers web interface becomes slower and slower until unusable, with a positive correlation between high network load.

When I posted the live graph firstly at the end of last week I was only on one upstream, but things improved when I got a 2nd added, and it was only about 3 weeks ago I went from 4 to 6 downstreams, so it would seem Virgin are still mid way through capacity upgrades in Stafford, which has never been a good area (high student population) even though FTTC has been available for nearly 2 years here now