#arpnetworks 2015-06-28,Sun

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
***dwarren has quit IRC (Read error: Connection reset by peer)
dwarren has joined #arpnetworks
[00:46]
.............. (idle for 1h9mn)
dwarren has quit IRC (Read error: Connection reset by peer)
dwarren has joined #arpnetworks
mnathani has quit IRC (Ping timeout: 265 seconds)
mnathani has joined #arpnetworks
[01:56]
..... (idle for 22mn)
kellytkAre there any conventional tiers of RTT latency? (eg., 0-30ms, >30-60ms, ..)
The only non-arbitrary basis for that that I can see at present is classes of applications based on latency, such as VOIP vs HTTP
https://en.wikipedia.org/wiki/Traffic_classification#Typical_traffic_classes is the nearest I've found, and it's based on logical application rather than latency, although that's implied
[02:19]
BryceBotTraffic classification :: Traffic classification is an automated process which categorises computer network traffic according to various parameters (for example, based on port number or protocol) into a number of traffic classes. Each resulting traffic class can be treated differently in order to differentiate the service implied for the user (data generator/ consumer). Typical uses Packets are classified to be differently processed by the. [02:22]
kellytkInteresting traffic introspection product read BTW: https://www.sandvine.com/downloads/general/sandvine-technology-showcases/traffic-classification-identifying-and-measuring-internet-traffic.pdf [02:26]
............. (idle for 1h2mn)
mercutiodon't think there is kelly..
jitter is usually referenced more than rtt..
as rtt can vary depending on distance
i used to disapprove of networks with more than 0.5 msec jitter in normal operation.
but there are just way too many factors..
[03:28]
kellytkThat's quite a tight tolerance [03:31]
mercutiobut in a very round about way, things like 2 msec jitter mean that there is more network load, which means the chances of upstream congestion issues in the near future is higher
yeah - things like virtualisation, coalescing etc can add a little bit of jitter which doesn't necessarily impact
[03:31]
kellytkI like how high jitter exposes providers with inadequate network capacity at certain times of a day.
It's also why I value throughput monitoring
[03:32]
mercutiothe thing is it's usual for jitter to go up a little as networks are getting into higher utilisation %'s.. which can suddenly lead to severe issues
well i like the idea of small file throughput monitoring still
[03:32]
kellytkI dislike over-subscription personally
That's what I mean
[03:32]
mercutiovultr sydney has had congestino issues since it started, but it's in the upload direction...
well it was in both directions for a while.
[03:32]
kellytkUp/down is a great distinction. I'll make a note of that [03:33]
mercutiobut it varies where it's from, a lot of weird issues, ... if you want somewhere weird to test from
err uploading to vultr is bad i mean
rather than downloading to it
err downloading from it
which suggests either people torrenting from it, or it sharing bandwidth with some isp work loads :)
the problem is most providers have more than one provider, and providers can have different issues in different locations
also it depends what you're doing, i stopped doing throughput monitoring mostly when 2 or 3 megabytes/sec was easy
because after that it stops mattering a lot more..
[03:33]
kellytkLatency, jitter, packetloss, throughput up/down. Can you think of other fundamental attributes of network performance? [03:36]
mercutioif you're getting 300k/sec it matters a lot more
mmomentary network outages
[03:36]
kellytkMy idea for that was to have 1MB file transfers frequently, with 10MB less frequently, and 100MB infrequently
Thoughts?
[03:37]
mercutiowell i do 200k
and i find it hilights any issues fine
and i do 10mb manually sometimes
[03:37]
BryceBotThat's what she said!! [03:37]
kellytkThat's good to know [03:37]
BryceBotThat's what she said!! [03:37]
mercutioi don't really think you need to do more than that
if it takes 2 seconds to download 200k when it usually takes 1 second there is degraded network performance
it doesn't really matter how fast you get now days
[03:37]
kellytkTrue [03:38]
mercutioit's more important to catch issues
and doing lots of 200k donwloads can still burn off 200gb of data ors omething
[03:38]
kellytkJitter is the standard deviation of a series of RTTs, correct? [03:40]
mercutiohttp://www.cisco.com/c/en/us/support/docs/voice/voice-quality/18902-jitter-packet-voice.html
actually that site sucks
jitter is basically the difference between median and mode i think, but don't hold me to that
[03:41]
kellytkIn plain terms, do you look at jitter as the standard deviation? To identify how much the latency varies [03:47]
............................................ (idle for 3h37mn)
***Hien_ has quit IRC (Quit: leaving)
Hien has joined #arpnetworks
Hien has quit IRC (Client Quit)
Hien has joined #arpnetworks
[07:24]
..................................................................................................................................................................... (idle for 13h42mn)
mnathanianyone know what it means when a thinkpad is stuck on battery light and z with circle indicator [21:10]
...... (idle for 27mn)
the screen turned to static like pixels the last time it was on.
I really hope it isn't dead
[21:37]
.... (idle for 16mn)
mercutiomaybe a short?
that's just a random guess though
[21:53]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)