Terrible performance Leicester (LE3) - Page 24

Chrysalis · 11-02-2011, 17:49

cpc14-leic14 I did wonder if I somehow been moved to your port but the graphs do have different trends still.

philce · 11-02-2011, 17:58

Yes very similar, but not exactly the same, must be some work ongoing.

Chrysalis · 12-02-2011, 00:21

I see the routing flapped again, I hope this isnt something local, as it created mayhem for my ssh sessions.

---------- Post added at 22:42 ---------- Previous post was at 22:39 ----------

looks like a flap on VM internal network, leeds.

---------- Post added at 23:21 ---------- Previous post was at 22:42 ----------

tons of packetloss, here is a inbound mtr during it. so between leics core and cmts, this will be fun reporting without it been ignored.

7. nrth-tmr-1-as0-0.network.virginm 0.0% 38 42.3 42.5 41.9 43.0 0.2
8. nrth-bb-1a-ae1-0.network.virginm 0.0% 38 45.2 54.7 44.6 189.7 28.9
9. nrth-bb-1b-ae0-0.network.virginm 0.0% 38 42.9 54.4 42.4 204.1 34.6
10. leic-core-1b-as0-0.network.virgi 0.0% 38 43.4 43.7 43.2 44.3 0.2
11. leic-cmts-14-gigaether-21.networ 22.2% 37 47.0 51.8 44.1 90.9 11.4
leic-cmts-14-gigaether-151.network.virginmedia.net
leic-cmts-14-gigaether-141.network.virginmedia.net
12. <xxxx> 22.2% 37 57.1 61.6 53.0 83.3 7.4

Chrysalis · 12-02-2011, 14:12

today jitter is notiebly higher than yesterday although today is a weekend, if I compare to phil's graph, yetserday my jitter was slightly lower than his but today its higher, so the trend I seen yesterday is probably not going to be the ongoing trend. It is still defenitly lower than before the outage but not such a huge gap now. In addition that 2nd upstream channel I was originally on appears to be completely inaccesible now.

qasdfdsaq · 12-02-2011, 14:16

I'm impressed by just how suboptimal the standard routing is if the route flaps can shave a third off your ping... Annoys me a bit somewhat too as it shows VM's network paths are at least able to perform on par with the competition, yet they choose for it to not do so.

Chrysalis · 12-02-2011, 14:29

yeah I seen that many times with isp's, they route where its cheapest to balance out utilisation on their network. VM's is suboptimal here most of the time it usually routes through leeds which adds about 6ms to the latency. Right now its not tho since late last night.

---------- Post added at 13:29 ---------- Previous post was at 13:25 ----------

I tested speedio again with ddwrt hooked up on double nat and is still full speed, so that wasnt to blame last time for that particular performance issue.

philce · 13-02-2011, 12:42

I dint get why there is such a difference in speedtests.

Speedtest.net is good for DL but UL is lower
Speedio is poor for Dl but good for UL.

Chrysalis · 13-02-2011, 15:34

there is 2 upstream channels now on my port, channel 7 (the new one) has very low latency and is easily lower utilised, it was very hard getting the modem onto it tho.

and daytime sunday.

we also in leics now seem to be on the lower latency routing since the flapping so 10-11ms (base latency) to bbc.co.uk at the moment.

speedio still ok here phil although I have wondered how reliable the transit is to that site after my initial issues with it. You able to do a speedtest to london sepeedtest server?

philce · 13-02-2011, 17:46

London looking good for Sunday PM, upload still a little low though, obviously local congestion?

qasdfdsaq · 13-02-2011, 17:55

It's not perfect but it shouldn't be causing real problems at those levels. Even for me web browsing and general internet use (and youtube at 1080p) still worked fine at ~600-800k upstream, though file downloads wouldn't go above ~25mb at that rate.

Chrysalis · 13-02-2011, 18:40

your graph looks bad today now phil so could be VM congestion.
apparently gloucs is also gigabit.

here is my today graph on channel7, which even tho much lower than channel8 is getting some jitter.

yesterday on channel8. typically sunday is higher usage than saturday.

philce · 13-02-2011, 18:59

Not looking too good from here though!

Yes congestion is increasing.

---------- Post added at 17:59 ---------- Previous post was at 17:58 ----------

On upstream channel 3 and down 129-132

Chrysalis · 14-02-2011, 19:26

since sunday morning I am happy with my upstream load, if it were to stay like this then no complaints, so if jitter goes up a little during peak I consider that acceptable.

today's.

only time will tell if a number of users get dumped on the port again.

philce · 14-02-2011, 20:46

Its amazing what happens when the kids come home from school!!

Not happy with the lack of response from the CEO's office. I emailed and called my contact, he hasn't even had the courtesy to call me back or email.

Very poor.

Chrysalis · 16-02-2011, 21:40

as far as my port is concerned the planned upgrades havent happened yet, still in march scheduled. So this change was unrelated and I am assuming is to do with uplift work. Relieved by that as jitter is already double of 3 days ago. VM really need to get on top of this and have tons of spare empty capacity for growth rather than upgrading so its only adequate and then getting completely overloaded a month later.